BEGIN:VCALENDAR VERSION:2.0 PRODID:-//AI Engineer//Code Summit 2026//EN CALSCALE:GREGORIAN METHOD:PUBLISH X-WR-CALNAME:AI Engineer Code Summit 2026 X-WR-TIMEZONE:America/Los_Angeles BEGIN:VEVENT UID:code-2026-session-dee568f7-812c-423e-96f0-8bce29ddaad2@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T160000Z DTEND:20261110T170000Z SUMMARY:Light Breakfast DESCRIPTION:Schedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 5 (Breakfast & Lunch)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:other STATUS:CONFIRMED TRANSP:TRANSPARENT END:VEVENT BEGIN:VEVENT UID:code-2026-session-627fbc54-d9e6-4366-95cd-2b70386a7ea3@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T160000Z DTEND:20261111T013000Z SUMMARY:Registration DESCRIPTION:Schedule: https://ai.engineer/code/2026/schedule LOCATION:Registration (outside Golden Gate)\, Hilton San Francisco Union Sq uare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:other STATUS:CONFIRMED TRANSP:TRANSPARENT END:VEVENT BEGIN:VEVENT UID:code-2026-session-1380080@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T165500Z DTEND:20261110T170000Z SUMMARY:MC Intro DESCRIPTION:Speakers: Richelle J.\n\nSchedule: https://ai.engineer/code/202 6/schedule LOCATION:Continental 1–3 (Workshop Room 1)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1302670@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T170000Z DTEND:20261110T172500Z SUMMARY:From Shared Language to Enforceable Contracts: Ontologies as Guardr ails for Agents. DESCRIPTION:Speakers: Frank Coyle\n\nThis talk echoes Pocock’s “shared understanding” and augments it -- turning that language into a machine- checkable model: typed entities\, relationships\, allowed states\, and inf erence rules.\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 1–3 (Workshop Room 1)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373448@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T170000Z DTEND:20261110T180000Z SUMMARY:Machine trust DESCRIPTION:Speakers: CJ Avilla\n\nAgents can now work for hours\, but most of us still approve every step or skip the prompts and hope. This session asks where we are on the curve for trusting agents\, and what has to be t rue before we hand them much more work. It draws on decades of trust resea rch\, from how societies learned to trust strangers to how pilots learned to trust the autopilot. Then we use Claude Managed Agents as a reference i mplementation for building systems that are safe\, secure and trustworthy by default.\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 4 (Workshop Room 2)\, Hilton San Francisco Union Squar e\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:workshop STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1376183@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T170000Z DTEND:20261110T180000Z SUMMARY:New Ways of Working with Codex DESCRIPTION:Speakers: Dominik Kundel\, Charlie Guo\n\nKeeping up with the r apid changes in coding agents can be exhausting. Join Dominik Kundel for a practical tour of recent features and workflows to bring you up to speed\ , without all the noise. We'll cover with live demos and walkthroughs how you can make the most out of your ChatGPT subscription whether you are pai r programming with Codex or having your dot manage your incidents.\n\nSche dule: https://ai.engineer/code/2026/schedule LOCATION:Continental 6 (Workshop Room 3)\, Hilton San Francisco Union Squar e\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:workshop STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1334020@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T170000Z DTEND:20261110T180000Z SUMMARY:Build a Coding Agent Harness From an Empty Loop DESCRIPTION:Speakers: Tejas Kumar\n\nStart with an agent that lies about fi nishing\, and leave with a harness you'd trust with your repo.\n\nA bare a gent loop ends when the model says it's done. In my AI Engineer Europe 202 6 talk\, that loop hit a login page and reported success anyway. The model wasn't the problem: nothing around it checked anything. In this workshop we fix that by hand\, one layer at a time\, in TypeScript.\n\nWe build eve ry part of an agent harness from scratch:\n\n- **The agent loop:** call th e model\, run the tools it asks for\, feed the results back.\n- **Guardrai ls:** iteration and context limits the model can't talk its way past.\n- * *Context management:** what stays in the window when a run gets long.\n- * *Deterministic verification:** read the trace\, run the linter and the tes ts\, and let a failed check overrule the model's claim.\n- **Retries and s ecrets:** a loop around the loop\, and credentials the model never sees.\n \nOne rule for the whole session: the prompt never changes. Every fix is c ode. You'll leave with a working harness on your laptop and the habit of f ixing the harness before the prompt. Starting repo: https://github.com/Tej asQ/basically-ai-harness\n\nSchedule: https://ai.engineer/code/2026/schedu le LOCATION:Continental 7–9 (Workshop Room 4)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:workshop STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1333169@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T173000Z DTEND:20261110T175500Z SUMMARY:One team's path to trusting AI generated code DESCRIPTION:Speakers: Ankit Jain\n\nOur team merges 4x more pull requests t han we did a year ago. Almost all of that code is AI-generated\, human ass isted. We tried scaling code review to match — more reviewers\, AI revie w tools\, stricter checklists — and none of it worked. What worked was s topping the fight over diffs entirely and building a verification layer ar ound *intent*.\n\nThis is the story of how one engineering team replaced t he PR review bottleneck with an intent-driven verification workflow: captu ring what a change *should* do before a single line is generated\, encodin g team standards as reusable invariants\, and letting deterministic checks — not tired reviewers — produce the evidence that a change is safe to ship.\n\nI'll walk through the specific practices that made the differenc e — especially how we turned years of recurring review comments into mac hine-enforceable invariants — and share the results\, the failures\, and what we'd do differently.\n\nSchedule: https://ai.engineer/code/2026/sche dule LOCATION:Continental 1–3 (Workshop Room 1)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1379134@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T181000Z DTEND:20261110T183500Z SUMMARY:The True Cost of AI Coding DESCRIPTION:Speakers: Scott Tolinski\n\nAfter seeing the mental health effe cts of AI coding throughout the industry\, Scott surveyed nearly 4000 deve lopers about how agentic workflows are affecting their mental health. Hear what's happening and why.\n\nSchedule: https://ai.engineer/code/2026/sche dule LOCATION:Continental 1–3 (Workshop Room 1)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1356771@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T181000Z DTEND:20261110T191000Z SUMMARY:Forget AGI. You're Surrounded by Low-Hanging Fruit. DESCRIPTION:Speakers: Lucas Meijer\n\nSoftware development lifecycles are f ull of slow processes you can automate. Let's fix those before we go prete nd to be AI astronauts.\n\nWe'll cover a range of easy-to-adopt techniques for that kind of automation\, and for pushing back against the agent-supe rcharged entropy that keeps trying to make your codebase slop out of contr ol.\n\nYou'll leave with prompts\, skills and workflows you can put to wor k the same afternoon.\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 4 (Workshop Room 2)\, Hilton San Francisco Union Squar e\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:workshop STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373449@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T181000Z DTEND:20261110T191000Z SUMMARY:Build a Multi-Agent PR Review and Repair Workflow with Omnigent DESCRIPTION:Speakers: Elise Gonzales\, Bradford Chang\n\nBuild a GitHub CI workflow that uses specialist agents to review pull requests\, coordinate bounded repairs\, and validate the resulting changes. In this hands-on wor kshop\, you’ll connect agents across models and coding harnesses\, defin e their roles and permissions\, and test the pipeline while Omnigent handl es session lifecycles\, delegation\, and messaging. We’ll draw on how Om nigent uses its own agents to manage incoming issues and contributions as an active open-source project\, examining what it takes to turn review fin dings into tested changes ready for human approval. You’ll leave with ve rsion-controlled agent configurations and a runnable workflow you can adap t to your own projects.\n\nSchedule: https://ai.engineer/code/2026/schedul e LOCATION:Continental 6 (Workshop Room 3)\, Hilton San Francisco Union Squar e\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:workshop STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373450@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T181000Z DTEND:20261110T191000Z SUMMARY:Stop Prompting\, Start Delegating: Hand off work to a team of agent s DESCRIPTION:Speakers: Ross Cefalu\n\nMost teams still use agents one task a t a time: prompt\, wait\, review\, repeat. In this hands-on workshop\, you 'll go further and hand an entire stream of work to agents with Factory. B ring your laptop. We'll take a real project from goal to merged PRs\, sett ing up the scope\, guardrails\, and review checkpoints that make delegatio n safe. You'll leave with a workstream running and a playbook for deciding what to hand off next.\n\nSchedule: https://ai.engineer/code/2026/schedul e LOCATION:Continental 7–9 (Workshop Room 4)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:workshop STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1357063@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T184000Z DTEND:20261110T190500Z SUMMARY:One Rule\, One Throwaway Skill: Agent Migrations with ast-grep DESCRIPTION:Speakers: Justin Poehnelt\n\nCoding agents struggle with large migrations. A migration produces thousands of diffs\, agents apply the sam e fix five different ways\, and nobody can reliably say how much work rema ins. I migrated a production codebase to Effect by pairing each change wit h an ast-grep rule that matches the old pattern and a temporary skill that tells an agent how to fix a match.\n\nThe list of matches told agents wha t to work on and told me how much was left. It also divided the work acros s parallel agents without overlap. But agents could also make a match disa ppear without fixing it\, so I had to write rules that were hard to evade. I’ll show how the rule and skill pair worked on a real migration and wh at I would change next time.\n\nSchedule: https://ai.engineer/code/2026/sc hedule LOCATION:Continental 1–3 (Workshop Room 1)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-a8ad909b-f03d-416b-99e7-591813ebb8e5@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T191000Z DTEND:20261110T194000Z SUMMARY:Networking Break DESCRIPTION:Schedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 5 (Breakfast & Lunch)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:other STATUS:CONFIRMED TRANSP:TRANSPARENT END:VEVENT BEGIN:VEVENT UID:code-2026-session-1334659@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T194000Z DTEND:20261110T200500Z SUMMARY:Inference Engineering for Coding Agents DESCRIPTION:Speakers: Philip Kiely\n\nCoding agents are a dominant use case for LLMs\, but they tend to use generalized inference setups even at larg e scale. Like any optimization problem\, in inference you can get better r esults with better constraints. This talk shares specific techniques that you can apply when building an inference service for exclusive use in codi ng agents\, including targeted quantization and pruning schemes\, domain-s pecific post-training\, KV cache management around compaction events\, sub agent session affinity in a world of autoscaling and load balancing\, and code-specific speculative decoding techniques across both common model-bas ed approaches (DFlash\, DSpark\, EAGLE) and deterministic speculation usin g n-grams.\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 1–3 (Workshop Room 1)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373451@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T194000Z DTEND:20261110T204000Z SUMMARY:Build Your Own Agent Harness with eve and Vercel DESCRIPTION:Speakers: Owen Kephart\n\nWhat does it take to build an agent t hat works across users\, tools\, and shared environments? In this hands-on workshop\, we’ll introduce eve\, Vercel’s agent framework\, and explo re the fundamentals of building your own agent harness.\n\nWe’ll walk th rough the architecture behind multiplayer agents and how Vercel’s infras tructure fits together: Workflows for durable execution\, Sandbox for isol ated code execution\, and AI Gateway for model access. Then we’ll put th ese concepts into practice through a guided exercise you can adapt to your own use case.\n\nYou’ll leave with a practical understanding of the pie ces behind an agent harness and a starting point for building and running your own agents on Vercel.\n\nBring your laptop for the hands-on portion.\ n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 4 (Workshop Room 2)\, Hilton San Francisco Union Squar e\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:workshop STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1381709@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T194000Z DTEND:20261110T204000Z SUMMARY:Ahmad Osman\, Founder & CEO\, Osmantic DESCRIPTION:Speakers: Ahmad Osman\n\nDetails TBD.\n\nSchedule: https://ai.e ngineer/code/2026/schedule LOCATION:Continental 6 (Workshop Room 3)\, Hilton San Francisco Union Squar e\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:workshop STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373453@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T194000Z DTEND:20261110T204000Z SUMMARY:Closing the Agentic Loop: Getting Agents to Safely Build\, Verify\, and Ship DESCRIPTION:Speakers: Kent C. Dodds\n\nAgents can write code. That's not th e hard part anymore. The hard part is trusting what they ship. In this han ds-on workshop\, I'll show you the loop I run every day: agents safely bui ld\, verify\, and deploy the outcomes I want\, and that loop is hooked up to signal (my own ideas\, user feedback\, bug reports\, and friction) so i t keeps itself going. I haven't read 1% of the code in my product\, and ag ents author and merge most of the PRs. That only works because of the syst em around them.\n\nWe'll start by running autopilot (npx kcd-skills@latest autopilot) on a real repo\, yours if you want\, and use the questions it asks to lay the foundation: invariants\, decision records that say when it 's OK to change them\, a primitives catalog\, and an AGENTS.md that's a ro uter\, not a novel. Then we'll get into verification. That means separate reviewer agents\, PR templates that have you review systems instead of cod e\, cloud agents with their own computers that show their work\, and a con trol CLI your agents maintain for themselves. Finally\, we'll limit the bl ast radius with feature flags\, health checks\, and backups you've actuall y restored\, and wire up the loops that keep it all moving. Bring a laptop and a repo you own.\n\nWhat you'll walk away with:\n- An autopilot plan f or your own repo\, with each step your agents can take on next\n- A primit ives catalog and invariants your agents actually follow\n- A PR template b uilt around intent\, one-way vs. two-way doors\, and system diagrams\, so you review systems instead of code\n- A verification setup: an AI reviewer \, agents that test the running app\, and a control CLI they maintain\n- A blast-radius checklist: flags with an owner\, a metric\, and a removal co ndition\, health checks after deploy\, and tested restores\n- The loops th at keep it going: bug reports\, user feedback\, friction\, and nightly cle anup\n\nPrerequisites: nothing to install ahead of time\, but a good inter net connection matters. To follow along on your own code\, bring a laptop with a coding agent you already use (Cursor\, Claude Code\, Codex\, etc.) signed in\, plus a GitHub repo you own and can open pull requests on.\n\nS chedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 7–9 (Workshop Room 4)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:workshop STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1381496@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T201000Z DTEND:20261110T203500Z SUMMARY:Proactive Agent Engineering DESCRIPTION:Speakers: Sydney Runkle\n\nWe’ve seen a shift in the architec ture of bleeding edge agents\, motivated by the design of popular personal agents like Muse and Instinct. The latest agents run in the background an d make progress on much larger goals. These proactive agents use primitive s like schedules and event triggers\, async subagents\, browser use\, and integrations with Slack and other messaging channels. Perhaps most importa ntly\, they all have robust memory systems\, which I’ll argue are the ke y to hitting the Goldilocks standard for proactivity.\n\nSchedule: https:/ /ai.engineer/code/2026/schedule LOCATION:Continental 1–3 (Workshop Room 1)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-29c87f1a-5570-4b60-9165-36e259a6aa07@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T204000Z DTEND:20261110T215000Z SUMMARY:Lunch DESCRIPTION:Schedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 5 (Breakfast & Lunch)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:other STATUS:CONFIRMED TRANSP:TRANSPARENT END:VEVENT BEGIN:VEVENT UID:code-2026-session-1341863@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T215000Z DTEND:20261110T221500Z SUMMARY:From Coding agents to Knowledge work agents DESCRIPTION:Speakers: Karan Vaidya\n\nHow harness evolved from code to Know ledge work\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 1–3 (Workshop Room 1)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373447@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T215000Z DTEND:20261110T225000Z SUMMARY:AFK Engineering: From Terminal to Voice-Controlled Remote Agents (P art 1) DESCRIPTION:Speakers: Philipp Schmid\, Thor Schaeff\, Ivan Leo\n\nYou will move from local coding in Google Antigravity to building a sandboxed remot e agent on Gemini Managed Agents with reusable skills\, vaulted credential s\, and remote MCP. From there\, you will connect the Gemini Live API to t rigger remote tasks and steer your agents by voice.\n\nSchedule: https://a i.engineer/code/2026/schedule LOCATION:Continental 4 (Workshop Room 2)\, Hilton San Francisco Union Squar e\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:workshop STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1376182@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T215000Z DTEND:20261110T225000Z SUMMARY:A Field Guide to Agent Shapes: From Long-Horizon to Open-Ended DESCRIPTION:Speakers: Yacine Mahdid\n\nAgents in 2026 split into a bunch of distinct shapes like externalized context with RLMs\, agents messaging ea ch other\, harnesses the agent rewrites itself\, sandboxes\, orchestrators with sub-agents\, and brute-force recombination.\n\nThis talk goes throug h what each one helps with\, then looks at three real deployments in indus tries/domains generally lagging in AI adoption that reused some of these p atterns.\n\nWe finish on where this is heading in 2027 with models that ed it their own context\, tweak other networks\, get trained for exploration earlier\, and learn continuously from tasks they generate themselves.\n\nS chedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 6 (Workshop Room 3)\, Hilton San Francisco Union Squar e\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:workshop STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1357839@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T215000Z DTEND:20261110T225000Z SUMMARY:Keep the Harness\, Own Your Inference DESCRIPTION:Speakers: Francisco Javier Arceo\n\nClaude Code without Claude. Codex without an OpenAI model.\n\nWe’ll put both harnesses to work on a real codebase with open-weight models running on a DGX Spark. They’ll i nspect the repository\, change code\, and test their work. The terminal ex perience stays familiar. The entire stack runs on hardware you control.\n\ nDrawing on our work building vLLM Agentic API\, we’ll show how we’re bringing familiar coding harnesses to inference engines and distributed se rving stacks including vLLM\, SGLang\, NVIDIA Dynamo\, and llm-d. The goal is to make the harness\, model\, and serving infrastructure independent c hoices.\n\nWith Codex Cloud\, the coding harness itself runs remotely. Beh ind that harness\, the Responses API adds another layer of agentic executi on: the server can manage conversations\, execute tools\, and retrieve con text. We’ll open up that machinery and explore how responsibility gets d ivided between the harness and the agentic server. \n\nDrawing on our work building vLLM Agentic API\, we’ll dig into conversation state\, memory through retrieval and file search\, and compaction: what the agent remembe rs\, how it finds what it needs\, and what survives when its context fills up. As more of the agent runs on servers\, owning your inference means un derstanding—and controlling—this execution stack.\n\nYou’ll leave wi th the setup and verification steps to run your own coding agent backend —from a DGX Spark on your desk to a shared GPU cluster. Choose the model \, choose the engine\, and scale the infrastructure while keeping the codi ng tools you already use.\n\n\nKey takeaways\n\n- Run it locally: connect Codex and Claude Code to open-weight inference on an NVIDIA DGX Spark.\n- Take it to a cluster: connect the same agent-facing API layer to distribut ed serving with llm-d or NVIDIA Dynamo.\n- Make the whole session work: pr eserve conversation history\, tool handoffs\, and streaming across backend s\, then verify the result through a completed coding task.\n\nSchedule: h ttps://ai.engineer/code/2026/schedule LOCATION:Continental 7–9 (Workshop Room 4)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:workshop STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1356697@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T222000Z DTEND:20261110T224500Z SUMMARY:The new job of the data team is providing context to AI DESCRIPTION:Speakers: George Fraser\n\nBusiness intelligence is dead\; long live context! The new job of every data team is to provide context to AI\ , so that when your head of sales asks “give me your 3 best ideas for im proving pipeline conversion”\, it gives 3 great answers grounded in accu rate\, correct queries. The tools for providing this solution are the same tools you know and love: AI is just a new consumer of the data stack. Fiv etran + dbt labs CEO George Fraser will take you through how real users so lve this problem today\, and demo live how to connect AI to real data.\n\n Schedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 1–3 (Workshop Room 1)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1352199@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T230000Z DTEND:20261110T232500Z SUMMARY:Building a Shared AI Assistant: The Platform Challenges Beyond MCP DESCRIPTION:Speakers: Utkarsh Shrivastava\n\nTwo years ago\, teams asked ou r platform team to connect their data to an AI assistant. Today they ship their own MCP servers in a week. That speed is a win\, and it swaps one en gineering problem for a harder one: how do you decide what gets through\, what gets gated\, and who owns the gates? This talk gives you a way to ans wer that — what stays team-owned\, what becomes shared platform\, and ho w to tell whether consolidating actually helped.\n\nAt Netflix\, bringing these tools together creates real value. Functions like Finance and Talent can now connect information that previously lived in separate systems. A business user can ask a question in one place instead of taking it to thre e teams. But that same reach creates ambiguity. Multiple tools expose diff erent versions of the same entity\, or claim to answer the same question. As the catalog grows\, the assistant has more ways to choose incorrectly. Sometimes a data-owning team discovers that another team has built on its data only when an evaluation fails.\n\nHow do you build a shared AI assist ant that brings tools and data from across a company into one place? At Ne tflix\, product teams increasingly build their own MCP integrations. Our c hallenge is making those capabilities work together so employees can get a nswers across teams through a single assistant.\n\nThis talk covers the de cisions and mechanisms behind that shift: connector certification\, an int ernal benchmark built from real employee queries that we use to evaluate r outing between overlapping tools\, and how we classify actions to separate what an agent can do on its own from what needs a person first. We'll exa mine where our initial domain boundaries broke down\, what required agreem ent between teams\, and why better tool descriptions alone couldn't resolv e the conflicts.\n\nWe'll close on what we haven't solved: who owns the au thoritative version of an entity\, how to retire a connector other teams d epend on\, and when the benefits of a shared capability justify the coordi nation it requires.\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 1–3 (Workshop Room 1)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373612@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T230000Z DTEND:20261111T000000Z SUMMARY:AFK Engineering: From Terminal to Voice-Controlled Remote Agents (P art 2) DESCRIPTION:Speakers: Philipp Schmid\, Thor Schaeff\, Ivan Leo\n\nYou will move from local coding in Google Antigravity to building a sandboxed remot e agent on Gemini Managed Agents with reusable skills\, vaulted credential s\, and remote MCP. From there\, you will connect the Gemini Live API to t rigger remote tasks and steer your agents by voice.\n\nSchedule: https://a i.engineer/code/2026/schedule LOCATION:Continental 4 (Workshop Room 2)\, Hilton San Francisco Union Squar e\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:workshop STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1357867@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T230000Z DTEND:20261111T000000Z SUMMARY:Hands-on Gitar: An AI Code Verification Agent for Your CI DESCRIPTION:Speakers: Taylor Luttrell-Williams\n\nGitar is Sonar’s AI cod e verification agent\, complementing SonarQube with an understanding of ea ch change’s context and intent. In this hands-on workshop\, you’ll ins tall Gitar as a GitHub App and use it across three pull request scenarios that show how verification goes beyond code review. Gitar verifies every c hange against your project's context and rules\, catches problems\, genera tes working fixes\, and iterates until your CI pipeline passes. You'll lea ve knowing how to set up Gitar on your own repositories. You'll also see h ow it delivers ready-to-merge\, green PRs with high signal and low noise\, while your team keeps control.\n\nSchedule: https://ai.engineer/code/2026 /schedule LOCATION:Continental 6 (Workshop Room 3)\, Hilton San Francisco Union Squar e\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:workshop STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373454@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T230000Z DTEND:20261111T000000Z SUMMARY:Build a personal agent with Exa Search DESCRIPTION:Speakers: Jakub Hojsan\n\nA personal agent gets asked everyday questions: where to eat near the office that's open late\, whether it will rain on Saturday\, how the Warriors did last night\, what a stock did thi s week. Exa answers all of these through one search call. Places come back with addresses\, hours and reviews\; weather comes back as daily forecast s\; sports come back as scores\, standings and schedules. You don't need a separate API for each. In this session\, Jakub Hojsan\, forward deployed engineer at Exa you show you how to build a personal agent on Exa search that handles a day's worth of real requests\, from finding dinner to plann ing around the weather. We will provide a repo to work out of as part of t his workshop.\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 7–9 (Workshop Room 4)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:workshop STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1334565@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261110T233000Z DTEND:20261110T235500Z SUMMARY:Building World-class Agentic GTM: Learnings from Monaco for Monaco DESCRIPTION:Speakers: Mihail Eric\n\nBuilding a custom agents platform is o ne of the key pieces to unlocking AI for complex verticalized business wor kflows. Monte Carlo is the sales-specific system we've built to help scale Monaco's revenue: a platform where revenue teams encode their own playboo ks as agents that run against everything we've ingested about their busine ss\, including pipeline\, outbound\, calls\, and the CRM. Sales is a parti cularly complex vertical for the following reasons: no two revenue orgs lo ok alike\, the data is large and heterogeneous\, and the expertise that ma tters lives in per-rep workflows nobody has written down. Over the last 2 years\, we’ve built a custom harness and orchestration system to power o ur world-class GTM team. \n\nIn this talk\, I'll break down the components a verticalized agent platform needs in order to be fast and correct\, inc luding an execution plane\, memory and curated implicit and explicit conte xt\, how expertise gets represented\, and the evaluation loop\, along with what is still unresolved.\n\nSchedule: https://ai.engineer/code/2026/sche dule LOCATION:Continental 1–3 (Workshop Room 1)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-3e1a3db8-9da0-4015-af37-7277b48182cb@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T000000Z DTEND:20261111T003000Z SUMMARY:Networking Break DESCRIPTION:Schedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 5 (Breakfast & Lunch)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:other STATUS:CONFIRMED TRANSP:TRANSPARENT END:VEVENT BEGIN:VEVENT UID:code-2026-session-1356850@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T003000Z DTEND:20261111T005500Z SUMMARY:Catching Reward Hacking From the Inside DESCRIPTION:Speakers: Daniel Balsam\n\nCongrats! Your agent passed every te st. But did it fix the bug in your code\, or did it just learn how to chea t its way through the tests? \nThe July incident where OpenAI’s agents e scaped their sandbox and breached Hugging Face made unaligned agentic beha vior a mainstream headline. Reward hacking\, eval awareness and cyber misu se are difficult to spot with black box methods\, like output filters and LLM judge monitors. As a result\, unaligned behavior can go undetected. \n At Goodfire\, we build monitors that read model’s internal activations. Interpretability lets us find representations tied to behaviors like rewar d hacking\, awareness of being evaluated\, and offensive cyber activity\, then build into probes that fire while the model is still generating a res ponse. That means we can sometimes catch a model heading toward bad behavi or before it even happens. In this talk\, I’ll walk through how we deplo y activation monitors production and share lessons from real deployments w ith frontier labs and inference providers. I’ll share a recent example o f how our monitors caught a coding agent building a seemingly innocuous te st program well before the agent ultimately used it to cheat\, and how our platform saves on compute by monitoring activations the model is already generating. I’ll also walk through how activation monitoring fits into a larger\, multi-layer defense system and how to set the right thresholds f or intervention.\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 1–3 (Workshop Room 1)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1304940@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T003000Z DTEND:20261111T013000Z SUMMARY:Our Agents Shipped 6\,000 PRs to 600+ Companies in The First Month. Now They Build Features DESCRIPTION:Speakers: Alex Lebedev\n\nPostHog Self-driving wrote 6\,000+ pu ll requests in its first month that 600+ companies chose to merge. We show ed the signal-to-PR pipeline at AI Engineer in London this April\, and rel eased it in late June. I'll show what happened since: where agents overrea ched\, what we threw away\, what's still science fiction\, and the jump fr om fixing bugs to building features. The system now pitches features backe d by your product data\, builds them\, and reviews them paranoidly\, testi ng like a real customer. All open source. Real PRs\, real reviews\, real f ailures\, shown live.\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 4 (Workshop Room 2)\, Hilton San Francisco Union Squar e\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:workshop STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1383277@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T003000Z DTEND:20261111T013000Z SUMMARY:Do it All with Devin DESCRIPTION:Speakers: Giovanni D'Antonio\n\nWhat is the future of post trai ning in coding. What a modern research team looks like\, and how Devin is part of it.\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 6 (Workshop Room 3)\, Hilton San Francisco Union Squar e\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:workshop STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1375631@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T003000Z DTEND:20261111T013000Z SUMMARY:Build your first Kiro Crew app: An agent that triages your backlog while you sleep DESCRIPTION:Speakers: Arundeep Nagaraj\, Dan Kiuna\n\nMost people meet AI a gents as a chat bot. In this hands-on workshop you will build one that wor ks while you are not watching. Starting from an empty directory\, you will install Kiro Crew\, an open-source orchestration layer for coding agents\ , and ship a working app: an agent that reads your repository’s open iss ues\, finds the duplicates\, explains why it thinks so\, and posts its fin dings on a schedule. Along the way you will give the agent persistent memo ry so it stops re-reading what it already knows\, fan the work out across background subagents so a thousand issues do not take a thousand turns\, a nd set the boundary that decides what it may and may not touch. Bring a la ptop and a GitHub repository you care about. You will leave with the app r unning\, and with a clear picture of what separates a demo from something you would actually let run unattended.\n\nSchedule: https://ai.engineer/co de/2026/schedule LOCATION:Continental 7–9 (Workshop Room 4)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:workshop STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1359774@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T010000Z DTEND:20261111T012500Z SUMMARY:Sandboxes are dead DESCRIPTION:Speakers: Vasek Mlejnsky\n\nEveryone and their mother has built a sandbox company. Good news: sandboxes are dead.\n\nWhy is the CEO of E2 B telling you sandboxes are dead? Shouldn’t he be telling you the opposi te? \n\nIn this talk\, you’ll learn why sandboxes took off\, exactly whe re they break down for agents\, and what I believe will replace them: a pr ivate cloud for every agent task that reshapes its infrastructure on the f ly to fit whatever the agent needs.\n\nYou’ll leave fully AI-pilled abou t the future of the cloud. Your agent is ready to keep going\, but your in frastructure is waiting for a human.\n\nSchedule: https://ai.engineer/code /2026/schedule LOCATION:Continental 1–3 (Workshop Room 1)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1380081@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T012500Z DTEND:20261111T013000Z SUMMARY:MC Outro DESCRIPTION:Speakers: Richelle J.\n\nSchedule: https://ai.engineer/code/202 6/schedule LOCATION:Continental 1–3 (Workshop Room 1)\, Hilton San Francisco Union S quare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-804ef31d-65f3-4d4f-8b16-25698ce1b28a@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T013000Z DTEND:20261111T043000Z SUMMARY:Opening Reception DESCRIPTION:Schedule: https://ai.engineer/code/2026/schedule LOCATION:Golden Gate (Expo Hall\, Lunch & Reception)\, Hilton San Francisco Union Square\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:other STATUS:CONFIRMED TRANSP:TRANSPARENT END:VEVENT BEGIN:VEVENT UID:code-2026-session-e8b1a15a-4a96-4ef0-92d7-99ac83615e72@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T160000Z DTEND:20261112T014000Z SUMMARY:Token Billionaires Lounge DESCRIPTION:Schedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 7 (Token Billionaires Lounge)\, Hilton San Francisco U nion Square\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:other STATUS:CONFIRMED TRANSP:TRANSPARENT END:VEVENT BEGIN:VEVENT UID:code-2026-session-f8b0a0a6-9f29-4434-92fe-694d66d20a34@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T160000Z DTEND:20261112T014000Z SUMMARY:Registration DESCRIPTION:Schedule: https://ai.engineer/code/2026/schedule LOCATION:Registration (outside Golden Gate)\, Hilton San Francisco Union Sq uare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:other STATUS:CONFIRMED TRANSP:TRANSPARENT END:VEVENT BEGIN:VEVENT UID:code-2026-session-1376240@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T170000Z DTEND:20261111T171000Z SUMMARY:MC Intro DESCRIPTION:Speakers: Corey Quinn\n\nSchedule: https://ai.engineer/code/202 6/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1352467@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T171000Z DTEND:20261111T173000Z SUMMARY:What Will You Build When AI Is 10x Faster? DESCRIPTION:Speakers: Natalie Serrino\n\nMany AI product decisions are infl uenced by how long someone will wait for an answer. Do you give an agent a nother pass to check its answer\, or make it respond faster? Can your voic e assistant finish reasoning before the pause gets awkward? Faster inferen ce gives developers the option to build smarter\, more responsive user exp eriences.\n\nWe’ll look at what faster inference lets you build\, from m ore responsive voice applications to agents that can try\, test\, and upda te within a few seconds. We’ll discuss when speed is worth paying for\, when other parts of your application become the bottleneck\, and how to us e higher responsiveness. Along the way\, we’ll share how Gimlet combines GPUs and specialized accelerators to make inference faster\, and what tha t means for the developers using that speed.\n\nSchedule: https://ai.engin eer/code/2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373455@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T173000Z DTEND:20261111T175000Z SUMMARY:Rethinking Claude Code Best Practices DESCRIPTION:Speakers: Lydia Hallie\n\nThe models have gotten really good\, and most of what we've added to our Claude Code setups over the past year is now holding them back. In this talk I'll go through what Claude Code be st practices look like for the Claude 5.5 models\, what's still worth keep ing\, and what you no longer need to think about.\n\nSchedule: https://ai. engineer/code/2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373457@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T175000Z DTEND:20261111T181000Z SUMMARY:cloudflare is a neolab DESCRIPTION:Speakers: Michelle Chen\n\nwe trained some new system one model s (clef\, clairic). they fit perfectly within our decade-old primitives li ke workers\, our edge latency hypothesis for workers ai\, and our ambition to be the agent cloud.\n\nSchedule: https://ai.engineer/code/2026/schedul e LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373446@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T181000Z DTEND:20261111T183000Z SUMMARY:All Code Is Slop (Except Mine) DESCRIPTION:Speakers: Lauren Tan\n\nThe Grok Bot team at SpaceXAI lands hun dreds of PRs a day. Nobody reads them all\, yet quality goes up anyway. In this talk\, we'll find out how the Grok Bot team turns slop into gold at an incredible pace.\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1376244@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T183000Z DTEND:20261111T184000Z SUMMARY:MC Outro DESCRIPTION:Speakers: Corey Quinn\n\nSchedule: https://ai.engineer/code/202 6/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-d4b20610-7cef-463f-b607-b5a51a334993@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T184000Z DTEND:20261112T000000Z SUMMARY:Expo Hall Hours DESCRIPTION:Schedule: https://ai.engineer/code/2026/schedule LOCATION:Golden Gate (Expo Hall\, Lunch & Reception)\, Hilton San Francisco Union Square\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:other STATUS:CONFIRMED TRANSP:TRANSPARENT END:VEVENT BEGIN:VEVENT UID:code-2026-session-a2b441c8-9b8e-489b-8da3-e6878f79cb02@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T184000Z DTEND:20261111T192000Z SUMMARY:Networking Break DESCRIPTION:Schedule: https://ai.engineer/code/2026/schedule LOCATION:Golden Gate (Expo Hall\, Lunch & Reception)\, Hilton San Francisco Union Square\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:other STATUS:CONFIRMED TRANSP:TRANSPARENT END:VEVENT BEGIN:VEVENT UID:code-2026-session-1380068@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T185500Z DTEND:20261111T191500Z SUMMARY:Manage Yourself Out of the Loop DESCRIPTION:Speakers: Erik Thorelli\n\nMore coding agents mostly means more tabs to watch. In the past year\, one of our engineers went from about 20 merged pull requests a month to nearly 7\,000. I'll show what we changed in our repos so an agent can take one small feature from prompt to merge o n its own.\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 1–3 (Expo Sessions)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1323755@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T192000Z DTEND:20261111T194000Z SUMMARY:I deleted 90% of my harness (And It Got Better) DESCRIPTION:Speakers: Philipp Schmid\n\nAlt Title: Why Models Are Superhum an with Bash (Where Humans Fail)\n---\nI spent my day to day building a ag ents with custom tools for everything. \n\nEvery few month I decide to del ete everything and see what is not needed anymore. I stripped out every de dicated tool until only Bash remained. I cut the prompt to 300 tokens and made the execution loop asynchronous.\n\nBenchmark scores jumped. Experien ce went up. \n\nRich Sutton called this the Bitter Lesson: building human heuristics into AI always loses to general methods backed by compute. Mode ls are now superhuman with the bash. A frontier model can synthesize an ad -hoc 15-line pipeline combining Python AST parsing\, ripgrep\, git diffs\, and subshells in 200ms to make progress on a users task\, there is no nee d for dedicated edits\, search or grep tools anymore. \n\nLearn how to del ete your code.\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373458@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T192000Z DTEND:20261111T194000Z SUMMARY:Refactor or Rot: Building a self-improving codebase DESCRIPTION:Speakers: Jacob Prall\n\nYour coding agents run in loops: plan\ , implement\, test\, repeat. This inner loop focuses on completing a task\ , and it tends to get all the attention. \n\nBut there’s an outer loop\, too. Each agent builds on the codebase the last one left behind\, inherit ing its patterns\, abstractions\, and shortcuts. Those patterns can make t he codebase easier to understand\, maintain\, and extend\, or compound com plexity and make future changes riskier.\n\nIn this talk\, I’ll show you how to build an outer loop that reinforces code quality and counteracts d ecay. Drawing on classic software design and refactoring techniques\, we ’ll equip an agent with skills to identify code smells\, reason about co upling and module boundaries\, and recommend refactors that make future ch anges easier. We’ll wire that analysis into the development lifecycle\, running on PRs and across the codebase\, weighing each recommendation’s benefit against its cost and risk.\n\nYou’ll learn when to refactor\, wh en to leave working code alone\, and how to make refactoring part of your development workflow\, so your coding agents leave behind codebases that a re better\, not just bigger.\n\nSchedule: https://ai.engineer/code/2026/sc hedule LOCATION:Continental 1–3 (Expo Sessions)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373459@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T194500Z DTEND:20261111T200500Z SUMMARY:The Rise of the Agent Workforce: Rethinking Agent Infrastructure DESCRIPTION:Speakers: Ankit Mathur\n\nThe next generation of agents won't j ust answer questions. They'll work autonomously\, collaborate with other a gents\, and learn from experience. As we move from interactive chatbots to persistent agent workforces\, the infrastructure powering them needs to f undamentally evolve. In this talk\, we'll explore the emerging architectur e of agentic systems\, from stateful sandboxes and persistent memory to ag ent communication and orchestration. We'll examine why today's agent platf orms aren't built for this future\, and what it will take to power a world where teams of agents can tackle complex\, long-running work together.\n\ nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373460@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T194500Z DTEND:20261111T200500Z SUMMARY:Heal Your Agents\, Don't Just Grade Them DESCRIPTION:Speakers: Khasimul Asif Mohammad\, Sudhir Tonse\n\nYour evals a re green\, and users are still hitting failures. The evals aren't wrong. T hey test what you anticipated\, and production at millions of sessions a d ay keeps inventing failures nobody wrote a test for\, many visible only as a pattern across sessions.\n\nDetecting those failures continuously is a hard problem in its own right. Every agent fails differently\, the failure s that matter are rare\, and a detector that flags everything is as useles s as one that flags nothing. We've been working on this with customers run ning agents at scale and by running it on our own agents\, learning which failures are worth surfacing and which are just noise.\n\nAlong the way\, three engineering problems kept coming up:\n\nEconomics: you can't afford to run a reasoning model over every session\, so what do you screen with\, and what earns a closer look?\n\nNoise: a thousand flagged sessions are u sually a handful of root causes. How do you collapse them and decide which to fix first?\n\nTrust: when a coding agent proposes the fix as a pull re quest\, what has to be true before you merge it?\n\nHow do you verify each fix and expand to cover every failure you find so it stays caught?\n\nYou 'll leave with the tradeoffs that matter when you move from grading agents to healing them\, drawn from teams doing it today\n\nSchedule: https://ai .engineer/code/2026/schedule LOCATION:Continental 1–3 (Expo Sessions)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373461@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T201000Z DTEND:20261111T203000Z SUMMARY:Building a Safe Software Factory DESCRIPTION:Speakers: Andrew Finch\n\nCoding agents have made code cheap to write\, and teams are shipping 10× more changes than they did two years ago. Human review was built for human speed\, so the industry is moving fr om reviewed PRs to automated software factories. At Gitar and Sonar\, we a re learning how to build a safe software factory\, with layers of verifica tion across the system. We'll share what we've been learned so far\, and w here we think this trend is headed.\n\nSchedule: https://ai.engineer/code/ 2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373462@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T201000Z DTEND:20261111T203000Z SUMMARY:How to build agents over your own data DESCRIPTION:Speakers: Aamir Shakir\n\nThe biggest bottleneck to building po werful internal agents is giving them the right organizational context. Wh en agents can reliably find that context\, they can answer questions\, rev iew code\, investigate incidents\, and automate workflows across the compa ny.\n\nBut company knowledge is scattered across Slack\, email\, Google Dr ive\, code repositories\, Linear\, internal documentation\, and multiple v ersions of the same information. Putting everything into the model’s con text is expensive\, risky\, and often reduces answer quality. Connecting e very source through separate MCP servers creates another bottleneck: the a gent must make numerous tool calls while inheriting each service’s laten cy\, rate limits\, and uneven search quality.\n\nThis talk shows how to se parate retrieval from reasoning by giving internal agents a unified search layer for company knowledge. We’ll explore how a search subagent breaks down complex tasks\, combines semantic and exact retrieval across systems \, preserves source permissions\, verifies freshness and versions\, and re turns compact\, cited evidence to the acting agent. Through realistic inte rnal workflows\, we’ll demonstrate how better retrieval makes agents mor e accurate\, traceable\, and useful while reducing tool calls\, latency\, and cost.\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 1–3 (Expo Sessions)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373463@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T203500Z DTEND:20261111T205500Z SUMMARY:Earning Autonomy: From Missions to Workstreams DESCRIPTION:Speakers: Luke Alvoeiro\n\nSeven months ago\, Factory introduce d Missions\, an orchestrator that ran worker agents through a full project with validation at every milestone. Real customer projects showed its lim its. It only handled code\, it was built for older models\, and its contex t belonged to one person instead of the team.\n\nLuke Alvoeiro explains ho w Factory rebuilt it. A more token-efficient Droid made running many agent s affordable. Environments let agents scale out inside a customer's own VP C. The new Missions gives each project a long-lived home with one orchestr ator\, shared context\, and more autonomy as validation earns trust.\n\nWh en a project moves into maintenance\, its mission graduates into a workstr eam: agents that turn bug reports\, alerts\, and ideas into validated\, sh ipped software.\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373464@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T203500Z DTEND:20261111T205500Z SUMMARY:The Software Factory That Needs Zero FDEs DESCRIPTION:Speakers: Raz Berman Gold\n\nWorking alone with Claude feels li ke a superpower. You open a session\, give it the right context\, and watc h it ship. Scaling that to a whole engineering org is where it breaks. One agent with more context\, more tools and more responsibility starts drift ing\, costs climb\, and nobody can tell which part failed. So most "AI sof tware factories" quietly ship with a team of forward-deployed engineers wh o wire up every customer by hand.\n\nWe built Overcut so a factory doesn't need them. Instead of one big agent\, it uses small agents with clear job s: triage\, root-cause analysis\, implementation\, code review and verific ation. Each one has its own context\, model and memory\, so it can be meas ured and improved on its own. Orchestration connects them. Workflows react to events in GitHub\, Jira and Slack\, run in parallel\, and stop at huma n approval gates where it matters. Because every piece is a reusable build ing block\, teams assemble their factory from ready-made parts instead of waiting for someone to build it for them.\n\nThen we'll prove it live. Sta rting from an empty project\, we'll build the factory on stage: a producti on alert in Slack becomes a Jira ticket with a root-cause analysis\, the t icket becomes a pull request\, and the PR is reviewed and fixed until it's ready to merge. No FDEs\, no hand-wiring.\n\nYou'll leave with architectu re patterns you can apply tomorrow \, whether you're one developer or roll ing out agents across a whole company.\n\nSchedule: https://ai.engineer/co de/2026/schedule LOCATION:Continental 1–3 (Expo Sessions)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-d3d5809a-6087-496f-8f90-9bb417a58d02@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T205500Z DTEND:20261111T220500Z SUMMARY:Lunch DESCRIPTION:Schedule: https://ai.engineer/code/2026/schedule LOCATION:Golden Gate (Expo Hall\, Lunch & Reception)\, Hilton San Francisco Union Square\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:other STATUS:CONFIRMED TRANSP:TRANSPARENT END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373465@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T220500Z DTEND:20261111T222300Z SUMMARY:I love you so jevving much DESCRIPTION:Speakers: Diogo Almeida\n\nLet's talk about the state of AI\, h ow something like jev stayed under the radar for years\, how we made it\, and most importantly: what comes next! I've been told that this crowd is t he absolute best of any conference for building the future\, so let's jam about sci-fi and finally build some sick smart software.\n\nSchedule: http s://ai.engineer/code/2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373466@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T220500Z DTEND:20261111T222500Z SUMMARY:Coding Agents and How to Tame Them DESCRIPTION:Speakers: Ivan Leo\n\nCoding agents are remarkably capable one minute and burning 50\,000 tokens in a broken loop the next. Over the past few months at Google DeepMind\, I've worked with Kaggle to audit over 50 coding benchmarks trace by trace and build a platform to evaluate frontier agents at scale. In this short 18-minute talk\, we'll cover the weird fai lure modes we caught in the wild\, the lessons we learned building air-gap ped sandboxes\, and how to take those same lessons to design better harnes ses\, verifiers\, and coding environments for your own agents.\n\nSchedule : https://ai.engineer/code/2026/schedule LOCATION:Continental 1–3 (Expo Sessions)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1374818@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T223000Z DTEND:20261111T224300Z SUMMARY:Beyond Generation: Fast\, Structured AI Decisions with Jev at Glean DESCRIPTION:Speakers: Chinmay Goyal\n\nNot every step in an AI product need s a full generative LLM call. Many are bounded decisions\, like understand ing what a user is asking for\, picking the best suggestion to show\, chec king whether an answer's citations hold up\, routing to the right tool or model\, or deciding if a request needs a guardrail. In this session\, I'll walk through how Glean has been exploring Jev across these decisions\, wh at we learned about where fast\, confidence-scored decisions shine\, and h ow we evaluate these models before they reach users.\n\nSchedule: https:// ai.engineer/code/2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1357875@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T223000Z DTEND:20261111T225000Z SUMMARY:More Checks\, Less Noise: Layered Security for AI-Assisted Developm ent DESCRIPTION:Speakers: Taylor Luttrell-Williams\, Killian Carlsen-Phelan\n\n More security checks should mean more confidence\, not more alerts to igno re. As AI accelerates code production\, how do we add independent verifica tion without turning development into an obstacle course? This session exp lores two dimensions of application security: using complementary analysis methods and placing checks at useful points throughout the software devel opment lifecycle. We’ll examine where deterministic analysis provides re peatable checks\, where probabilistic\, AI-powered review can add context- sensitive reasoning\, and why simply stacking tools can create more noise than assurance.\n\nWe’ll start with a vendor-neutral framework for choos ing checks\, prioritizing findings\, and keeping review actionable. Then\, a live demo will bring three complementary approaches together: static ap plication security testing (SAST)\, software composition analysis (SCA)\, and AI-powered analysis of logic-level vulnerabilities. You’ll leave wit h a practical way to evaluate layered verification: what each check contri butes\, where it belongs\, and how to turn its findings into manageable wo rk rather than another queue of alerts.\n\nSchedule: https://ai.engineer/c ode/2026/schedule LOCATION:Continental 1–3 (Expo Sessions)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1374822@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T224800Z DTEND:20261111T230100Z SUMMARY:Your Automation Is Overthinking It DESCRIPTION:Speakers: Chibi Vikram\, Kevin Bernstein\n\nEnterprise automati on gets hard to maintain when a single agent is responsible for understand ing a request\, finding evidence\, and deciding what happens next. Drawing on two decades of experience automating business processes\, we'll show h ow to break that work into steps a team can understand\, test\, and change . We'll explore when a focused judgment is enough\, when a generative mode l should investigate\, and when a person should step in. We'll share what we learned about accuracy\, speed\, and cost so you can make those choices in your own workflows.\n\nSchedule: https://ai.engineer/code/2026/schedul e LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373467@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T225500Z DTEND:20261111T231500Z SUMMARY:Ground Truth or Dare: Using Oracles to Make Postgres Reactive DESCRIPTION:Speakers: James Arthur\n\nLive SQL is the holy grail of realtim e. Subscribe to any SQL query\, with joins\, aggregates and ordering\, and the result stays live as your data changes. It’s easy to describe but v ery hard to build. At Electric\, we spent years chasing it. Then we joined Neon and built it in weeks with coding agents\, inside Postgres.\n\nLive queries are database infrastructure. This isn’t code you hand to an agen t and hope for the best. So we used Postgres’s own query engine as an or acle. Once the agents had the ability to verify their own work\, they coul d go super fast and burn down through the featue requirements.\n\nIn this talk\, I’ll cover what we built\, how we built it and how oracles let us have both truth and dare: to get it right and ship fast with agents.\n\nS chedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 1–3 (Expo Sessions)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1376197@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T230600Z DTEND:20261111T231900Z SUMMARY:Your Judge Is Hallucinating Too DESCRIPTION:Speakers: Brian Kennedy\n\nEvery step an agent takes in ZoomInf o's Agent Teams is graded by another model. Then we ran Jev beside it and found the grader was passing work that never happened. This talk is about what we found\, what we replaced\, and why "did the agent do a good job?" is the wrong question to ask a model.\n\nSchedule: https://ai.engineer/cod e/2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373468@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T232000Z DTEND:20261111T234000Z SUMMARY:CVE remediation with Factory: from finding to merged PR DESCRIPTION:Speakers: Nikhil Harithas\n\nDescription TBD\n\nSchedule: https ://ai.engineer/code/2026/schedule LOCATION:Continental 1–3 (Expo Sessions)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1376198@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T232400Z DTEND:20261111T233700Z SUMMARY:Scott Roberts\, Workday DESCRIPTION:Speakers: Scott Roberts\n\nSchedule: https://ai.engineer/code/2 026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1380927@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T234500Z DTEND:20261112T000500Z SUMMARY:Leaving Localhost: Building an Agent-Native Dev Environment DESCRIPTION:Speakers: Sajid Mehmood\n\nWhy is it that Anthropic can port Bu n from Zig to Rust in 11 days\, while most teams still struggle to make sw eeping changes to production software? Bun is complex\, but it turns out i ts verification loop is actually quite clear: there's a well-defined TypeS cript spec\, and a huge battery of tests the agent can iterate against.\n\ nThis talk will explore techniques to build agent-native dev environments in the cloud that allow for these kinds of tight verification loops for so ftware that's traditionally much harder to verify: tax accounting\, vertic al AI agents\, robotics control systems\, etc. We'll discuss how to manage stateful dependencies (databases\, queues\, caches)\, third-party service s\, and seed data so you can focus on architecture & product while your ag ents handle implementation & testing in the cloud.\n\nSchedule: https://ai .engineer/code/2026/schedule LOCATION:Continental 1–3 (Expo Sessions)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-222dd686-32de-4bf1-81fe-53ca33a74d1b@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261111T234500Z DTEND:20261112T000500Z SUMMARY:Afternoon Break / Keynote Reset DESCRIPTION:Schedule: https://ai.engineer/code/2026/schedule LOCATION:Golden Gate (Expo Hall\, Lunch & Reception)\, Hilton San Francisco Union Square\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:other STATUS:CONFIRMED TRANSP:TRANSPARENT END:VEVENT BEGIN:VEVENT UID:code-2026-session-1376245@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T002000Z DTEND:20261112T003000Z SUMMARY:MC Intro DESCRIPTION:Speakers: Corey Quinn\n\nSchedule: https://ai.engineer/code/202 6/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373469@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T003000Z DTEND:20261112T005000Z SUMMARY:What it takes to set intelligence free DESCRIPTION:Speakers: Fabian Hedin\n\nModels keep getting better and the sy stems we build around them need to keep up. This talk goes inside Lovable ’s engineering: forkable computers that give agents room to experiment\, persistent histories that let them continue longer work\, and durable com munication that lets them coordinate. We’ll explore what it takes to giv e agents more independence\, and how these foundations could help them bui ld\, ship\, learn from feedback\, and keep improving a product.\n\nSchedul e: https://ai.engineer/code/2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373470@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T005000Z DTEND:20261112T011000Z SUMMARY:How to build an agent that builds the culture DESCRIPTION:Speakers: Zach Bruggeman\n\nRamp built its own coding agent\, I nspect\, which writes ~75% of merged PRs. At the start of 2026\, the techn ology was the novel part\, but today\, it's commonplace. In this talk\, Za ch will talk about how Inspect has evolved this year\, and re-shaped the c ulture of engineering at Ramp around agent-driven development.\n\nSchedule : https://ai.engineer/code/2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373471@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T011000Z DTEND:20261112T013000Z SUMMARY:Steve Yegge\, Independent Advisor & Creator of Beads & Gas Town\, I ndependent DESCRIPTION:Speakers: Steve Yegge\n\nSchedule: https://ai.engineer/code/202 6/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1376191@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T013000Z DTEND:20261112T015000Z SUMMARY:Simon Willison\, Creator of Datasette\, Independent DESCRIPTION:Speakers: Simon Willison\n\nSchedule: https://ai.engineer/code/ 2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1376246@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T015000Z DTEND:20261112T020000Z SUMMARY:MC Outro DESCRIPTION:Speakers: Corey Quinn\n\nSchedule: https://ai.engineer/code/202 6/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-2f9774d5-ce23-415d-8047-2ffd94f4ee61@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T160000Z DTEND:20261113T014000Z SUMMARY:Token Billionaires Lounge DESCRIPTION:Schedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 7 (Token Billionaires Lounge)\, Hilton San Francisco U nion Square\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:other STATUS:CONFIRMED TRANSP:TRANSPARENT END:VEVENT BEGIN:VEVENT UID:code-2026-session-57c167d1-4feb-4c9d-aff2-94d59b9f831c@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T160000Z DTEND:20261113T014000Z SUMMARY:Registration DESCRIPTION:Schedule: https://ai.engineer/code/2026/schedule LOCATION:Registration (outside Golden Gate)\, Hilton San Francisco Union Sq uare\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:other STATUS:CONFIRMED TRANSP:TRANSPARENT END:VEVENT BEGIN:VEVENT UID:code-2026-session-1380094@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T170000Z DTEND:20261112T171000Z SUMMARY:MC Intro DESCRIPTION:Speakers: Corey Quinn\n\nSchedule: https://ai.engineer/code/202 6/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1379009@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T171000Z DTEND:20261112T173000Z SUMMARY:Jason Liu\, Developer Experience Engineer\, OpenAI DESCRIPTION:Speakers: Jason Liu\n\nSchedule: https://ai.engineer/code/2026/ schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373472@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T173000Z DTEND:20261112T175000Z SUMMARY:Armin Ronacher\, CTO\, Earendil DESCRIPTION:Speakers: Armin Ronacher\n\nSchedule: https://ai.engineer/code/ 2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373473@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T175000Z DTEND:20261112T181000Z SUMMARY:Software Engineering Is Becoming Factory Engineering DESCRIPTION:Speakers: Zach Lloyd\n\nWarp founder and CEO Zach Lloyd explore s the rise of software factories: systems that coordinate AI agents across the entire development lifecycle\, from triage and specification to imple mentation\, review\, and monitoring. He’ll explain how engineering teams can build factories that improve with every run\, where human judgment re mains essential\, and why the next frontier of software engineering may be less about writing code and more about engineering the systems that build it.\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1378217@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T181000Z DTEND:20261112T183000Z SUMMARY:Let's Open Source the World DESCRIPTION:Speakers: Brandon Thomas\n\nFrom open-source alternatives to Ad obe Photoshop and Microsoft Office to the SaaSpocalypse\, it’s clear tha t software’s moats aren’t where they used to be. Take a look inside Ar tCraft\, a project aiming to become the Wikipedia of open source by recrea ting every category of software in modern\, high-quality Rust. We’re bui lding open-source applications and infrastructure to make every platform o pen - tearing down moats by turning platforms into protocols. We’ll tour fully agentic software factories and the engineering behind them - pushin g the limits of Git and GitHub\, tackling Rust build and CI bottlenecks\, automating triage and testing\, and building high-quality shared libraries and UI/UX components for the interfaces and user experiences of today and tomorrow\, and beyond.\n\nSchedule: https://ai.engineer/code/2026/schedul e LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1380095@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T183000Z DTEND:20261112T184000Z SUMMARY:MC Outro DESCRIPTION:Speakers: Corey Quinn\n\nSchedule: https://ai.engineer/code/202 6/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-76f554f5-942f-4a36-86c4-09ed972006fc@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T184000Z DTEND:20261113T000000Z SUMMARY:Expo Hall Hours DESCRIPTION:Schedule: https://ai.engineer/code/2026/schedule LOCATION:Golden Gate (Expo Hall\, Lunch & Reception)\, Hilton San Francisco Union Square\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:other STATUS:CONFIRMED TRANSP:TRANSPARENT END:VEVENT BEGIN:VEVENT UID:code-2026-session-0ac34d11-ab5e-4861-88ce-4bbde76dfb23@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T184000Z DTEND:20261112T192000Z SUMMARY:Networking Break DESCRIPTION:Schedule: https://ai.engineer/code/2026/schedule LOCATION:Golden Gate (Expo Hall\, Lunch & Reception)\, Hilton San Francisco Union Square\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:other STATUS:CONFIRMED TRANSP:TRANSPARENT END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373475@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T192000Z DTEND:20261112T194000Z SUMMARY:Walden Yan\, Chief Product Officer\, Cognition DESCRIPTION:Speakers: Walden Yan\n\nSchedule: https://ai.engineer/code/2026 /schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373476@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T192000Z DTEND:20261112T194000Z SUMMARY:Emitted\, not authored: The context layer under your software facto ry DESCRIPTION:Speakers: Brandon Waselnuk\n\nIt's your first week on the team. You read everything attached to your first ticket: the README\, the wiki\ , the Linear project. You ship the PR. The review comes back from a senior engineer: "we stopped doing it this way in March\, the vendor doesn't acc ept that shape anymore." Nobody wrote that down. It lived in a Slack threa d from an incident two months ago and in the heads of the people who were up that night fixing it. You learned it by getting it wrong.\n\nThat's how most engineering knowledge gets earned. Not from documentation someone sa t down to write\, but from the act of the work itself: the pull request co mments\, the chat thread where a feature got argued out\, the incident cha nnel while people are still responding\, the support channel where a custo mer explained what they actually needed.\n\nI call this emitted context. I t was just typed\, by people\, doing their jobs.\n\nAuthored context is wh at the team decides matters: the rules and skills files for agents\, the w ikis and READMEs we curate. High leverage\, worth doing well. Emitted cont ext is what the team told each other while doing the work. Your agents tod ay are you\, the new hire\, on every single task\, and they never get to b uild tribal knowledge.\n\nLetting them build it is a lot harder than conne cting a Slack MCP. Emitted context contradicts itself\, goes stale overnig ht\, and includes things not everyone is allowed to see. This talk is abou t the context layer that handles that: surfacing context nobody authored\, reconciling it with citations\, keeping it fresh\, with no human in the l oop. Automate a factory on authored context alone and it can only act on w hat somebody remembered to write down\, and no one person knows everything the team knows.\n\nI'll run the same task with and without emitted contex t\, show why agents don't stop to ask for what they're missing\, and cover the three things we got wrong building this. You'll leave with open sourc e tools we extracted from our context layer\, and a clear read on whether your agents can unblock themselves or if you're trapped in the factory.\n\ nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 1–3 (Expo Sessions)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373477@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T194500Z DTEND:20261112T200500Z SUMMARY:What Your AI Doesn't Know: The latest trends in agent searches DESCRIPTION:Speakers: Jeffrey Pinner\n\nAn agent only searches when it does n’t know something. In other words\, every search is a confession\, and Exa has been collecting them from thousands of agents for two years.\n\nRe ad together\, those searches form a map of how AI agents like to work\, an d where their knowledge ends. There’s clear evidence of growing expertis e. Coding agents used to search for examples and tutorials\, now they go s traight to the source\, searching for code repos\, issues and API referenc es. Searches for mature frameworks like React\, Next.js\, and the rest of the “vibe code stack” are fading\, not because they’re falling out o f use\, but because models already know them cold. Instead agents are sear ching for their own tools: MCPs\, skills\, agent instruction files and the latest models.\n\nBut even as their knowledge and capability grows\, agen ts still need to search. Models still take months to train\, but the codin g frontier now turns over every quarter\, and AI is speeding that up. This knowledge cutoff won’t be fixed by the next big model release.\n\nIn th is keynote\, Jeff Pinner\, CTO of Exa\, shares what we’ve learned from a gents telling us what they don’t know\, what it means for anyone who mai ntains a library\, ships an agent or relies on one\, and why search is an Agent’s best gateway to the present.\n\nSchedule: https://ai.engineer/co de/2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1379052@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T194500Z DTEND:20261112T200500Z SUMMARY:Talk But Verify: human-in-the-loop for agents that actually matter DESCRIPTION:Speakers: Michael Liendo\n\nAI agents are getting good at takin g action. The hard part isn't making them capable\, it's knowing when to m ake them stop. In this talk we'll walk through the full spectrum of human- in-the-loop patterns\, from lightweight inline confirmations to out-of-ban d permission gates to handing your agent a wallet with real money in it an d more. Each pattern fits a different level of consequence\, and knowing w hich to reach for is what separates demo agents from production ones. We'l l cover the honest tradeoffs of latency\, user experience\, and trust so y ou can make the right call for your specific use case. The entire talk is built around various live demos that escalate in stakes with every step. Y ou'll leave with a mental model and working reference architecture you can apply the same day.\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 1–3 (Expo Sessions)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1375634@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T201000Z DTEND:20261112T203000Z SUMMARY:Ship more as a team of one with Kiro Crew DESCRIPTION:Speakers: Doug Clauson\n\nCoding has gotten cheaper\, but shipp ing hasn't. See how one engineer built a build\, review\, and ship loop wi th Kiro Crew\, and learn the traps that cost hours along the way (reviews that never end\, agent loops that balloon a ten-line fix into a whole subs ystem\, saying yes to everything because code felt free). You'll leave wit h five concrete habits to try Monday morning with the tools you already us e.\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1379011@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T201000Z DTEND:20261112T203000Z SUMMARY:Let the Agent Cook: From APIs to Answers DESCRIPTION:Speakers: Arsalan Bashir\n\nAbstract: Give an agent a pile of A PIs. Can it figure out what to query\, connect the dots\, and check its ow n answers? I’ll share practical lessons on building agents that explore data across services—from discovering tools and navigating schemas to fo llowing leads and knowing when they have enough evidence.\n\nSchedule: htt ps://ai.engineer/code/2026/schedule LOCATION:Continental 1–3 (Expo Sessions)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1379010@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T203500Z DTEND:20261112T205500Z SUMMARY:Exploring Claude’s frontier DESCRIPTION:Speakers: Lance Martin\n\nIn this talk\, we explore Claude’s frontier capabilities in memory and multi-agent coordination.\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1379163@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T203500Z DTEND:20261112T205500Z SUMMARY:Code review without reviewing code DESCRIPTION:Speakers: Daniel Tao\n\nWhile authoring code has become virtual ly instantaneous\, reviewing and verifying that code in an enterprise envi ronment remains constrained by human cognitive capacity. At scale\, this b ecomes a painful bottleneck: pull request volume and size are growing rapi dly\, review latency is increasing\, and 10\,000-line diffs can all too ea sily burn through finite reserves of human attention.\n\nHow do engineerin g organizations scale trust and quality at the speed of AI generation?\n\n I won’t pretend we’ve fully slain this dragon\, but in this session I ’ll walk through a few different approaches we’re taking and what we ’ve learned within Atlassian:\n\nInner Loop: Building confidence through automated local AI review and skills for catching anti-patterns and contr act violations before code ever reaches a human teammate.\n\nOuter Loop: H ow tools like Taro spin up live environments\, simulate user interactions\ , and provide tangible evidence (narrated videos) directly to reviewers.\n \nA New Lens on Code: Aggregating quality signals like CI\, agentic evalua tions\, and risk scoring to allow reviewers to “zoom out” and reason a bout the value and implications of a change without having to read the dif f line-by-line.\n\nRethinking the Unit of Change: Why the traditional sing le-ticket micro-PR is an artifact of human typing limits\, and how we can shift human expertise to where it matters most.\n\nAttendees will leave wi th concrete architectural patterns\, lessons learned from real-world rollo uts\, and a blueprint for evolving their team’s code review practices fo r the AI era.\n\nSchedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 1–3 (Expo Sessions)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-0d41a8b7-3e4d-4487-895e-c3330d82620e@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T205500Z DTEND:20261112T220500Z SUMMARY:Lunch DESCRIPTION:Schedule: https://ai.engineer/code/2026/schedule LOCATION:Golden Gate (Expo Hall\, Lunch & Reception)\, Hilton San Francisco Union Square\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:other STATUS:CONFIRMED TRANSP:TRANSPARENT END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373478@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T220500Z DTEND:20261112T225000Z SUMMARY:Jarred Sumner\, Founder of Bun\, Anthropic / Bun DESCRIPTION:Speakers: Jarred Sumner\n\nSchedule: https://ai.engineer/code/2 026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1380658@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T220500Z DTEND:20261112T222500Z SUMMARY:Scaling Your Agent Workforce: Lessons from Inside Cognition DESCRIPTION:Speakers: Theodor Marcu\n\nMost companies have turned coding ag ents on. Few can say what those agents do or what they return. At Cognitio n\, Devin wrote 92% of merged PRs in September\, and 9 out of 10 Devin ses sions start without a human. Theodor Marcu shows how we got there: cloud a gents working in parallel\, automations that run every night\, and how eng ineering and research use Devin to autonomously drive feature development. He then turns that into a playbook for your teams: govern what agents can reach\, route work across models to cut cost\, and follow every session t o a merged PR so ROI is measured instead of guessed.\n\nSchedule: https:// ai.engineer/code/2026/schedule LOCATION:Continental 1–3 (Expo Sessions)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1381580@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T223000Z DTEND:20261112T225000Z SUMMARY:The first Grounded Decision Model DESCRIPTION:Speakers: Devin Gupta\n\nDecision models only read. They never generate a token\, which is why they cost roughly 240x less per token read than a frontier LLM — and why they are only as good as the state you ha nd them. For most judgments worth making\, that state lives on the live we b\, not in your database.\n\nThis talk re-costs the stack with measured nu mbers. On PWS-Decisions\, our open 300-question benchmark\, the question i s 104 tokens and the evidence needed to answer it is 4\,777. Two years ago the model was 86% of the bill for a grounded answer\; today it is 17%. In telligence stopped being the bottleneck. Context became it.\n\nWe will cov er what that inversion changes architecturally\, a live demo that grounds 18 typed questions against 4\,270 businesses in a single call\, and the en tity-resolution failure that quietly breaks grounded decisions in producti on. You will leave able to run the benchmark against your own model.\n\nSc hedule: https://ai.engineer/code/2026/schedule LOCATION:Continental 1–3 (Expo Sessions)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1376195@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T225500Z DTEND:20261112T231500Z SUMMARY:Ishaan Dey\, Software Engineer\, Google DeepMind (Antigravity) DESCRIPTION:Speakers: Ishaan Dey\n\nSchedule: https://ai.engineer/code/2026 /schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373479@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T225500Z DTEND:20261112T231500Z SUMMARY:Your agent pays twice for every search: how we cut token spend by 9 0+% with better accuracy DESCRIPTION:Speakers: Louis Walsh\n\nEvery agent that searches the web pays twice: once for the search call\, and again for the tokens that search pu ts into the model's context. The answer is usually a few sentences. The ag ent pays for the whole page\, and in a multi-step loop it pays for it agai n on every turn.\n\nWe'll show real production traces of where the tokens go and why retrieval\, not generation\, is the biggest line on most agent bills. Then we'll show what we changed: return the passages a query needs instead of whole pages\, tune extraction to the question\, and cap how muc h context the agent can take. We tested this across 12 frontier models and 11\,000 blind-graded answers. Token spend fell by more than 90% while acc uracy went up.\n\nYou get a metric (cost per correct answer\, not cost per request)\, an open calculator to run the numbers on your own traffic\, an d a five-point checklist for finding token waste in your pipeline.\n\nSche dule: https://ai.engineer/code/2026/schedule LOCATION:Continental 1–3 (Expo Sessions)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373480@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T232000Z DTEND:20261112T234000Z SUMMARY:Brownfield Factories DESCRIPTION:Speakers: Justin McCarthy\n\nBrownfield applications are clearl y ideal for Software Factories! Brownfield applications provide all the sp ecification and telemetry you need to build validation harnesses and loops . Make a lump sum payment on your technical debt\, reduce costs\, increase performance\, and purge vulnerabilities once and for all. Rewrite that bo at anchor to Rust\, Elixir or BF today!\n\nSchedule: https://ai.engineer/c ode/2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1375635@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T232000Z DTEND:20261112T234000Z SUMMARY:Agents That Don't Clock Out DESCRIPTION:Speakers: Erik Hanchett\n\nMy coding agents are fast\, but spee d isn't the whole story. The real question is where they run: the CLI\, th e IDE\, or something else entirely? In this session\, I'll break down whic h tool fits which job and why that choice makes or breaks your workflow. I 'll show you how I organize skills and custom agents so they're reusable i nstead of one-off\, then close with a demo of a persistent agent setup in Kiro Crew — one that keeps grinding away long after I've logged off. Wal k away knowing how to match the right agent to the right environment and b uild setups that work while you sleep.\n\nSchedule: https://ai.engineer/co de/2026/schedule LOCATION:Continental 1–3 (Expo Sessions)\, Hilton San Francisco Union Squ are\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:breakout STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-066d8991-0108-4be6-8b75-6f423e76bafa@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261112T234500Z DTEND:20261113T000500Z SUMMARY:Afternoon Break / Keynote Reset DESCRIPTION:Schedule: https://ai.engineer/code/2026/schedule LOCATION:Golden Gate (Expo Hall\, Lunch & Reception)\, Hilton San Francisco Union Square\, 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:other STATUS:CONFIRMED TRANSP:TRANSPARENT END:VEVENT BEGIN:VEVENT UID:code-2026-session-1380097@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261113T002000Z DTEND:20261113T003000Z SUMMARY:MC Intro DESCRIPTION:Speakers: Corey Quinn\n\nSchedule: https://ai.engineer/code/202 6/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373474@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261113T003000Z DTEND:20261113T005000Z SUMMARY:Post-Training Gemini 4 for Real Engineering Work DESCRIPTION:Speakers: Melvin Johnson\n\nReal software engineering tasks are not asking a model to write a function\, but long tasks (o(hours) reading unfamiliar code\, running tools\, recovering from wrong turns\, and knowi ng when the job is actually done. In this keynote\, Melvin Johnson (VP\, G oogle DeepMind) covers how we post-trained for long-horizon agentic coding :\n\n- Hard tasks. Popular benchmarks saturate\, so we built internal real istic\, multi-step software tasks.\n\n- Graders that can't be gamed. When agents learn to satisfy unit tests without solving the problem\, you need graders that check the work the way a reviewer would.\n\n- Model behavior. An agent that says "fixed it" when it didn't is worse than no agent. We r eward models for reporting accurately what they did and didn't do.\n\n- Ef ficiency over long horizons. Finishing more of the task with fewer tokens and stable behavior across very long contexts.\n\n- Shipping weekly. Modul ar improvements and dogfooding with Google engineers let us release in wee ks instead of quarters.\n\nSchedule: https://ai.engineer/code/2026/schedul e LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373481@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261113T005000Z DTEND:20261113T011000Z SUMMARY:Quinn Slack\, CEO\, Amp DESCRIPTION:Speakers: Quinn Slack\n\nSchedule: https://ai.engineer/code/202 6/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1325061@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261113T011000Z DTEND:20261113T013000Z SUMMARY:12-Factor-Factories: Design Patterns for Agentic Coding DESCRIPTION:Speakers: Dexter Horthy\n\nEvery AI lab is building full stack cloud agents. Every enterprise is building their own software factory. We' ve worked with a lot of teams adopting all flavors in every combination. I n this talk we'll distill out the core principles and design patterns we'v e seen across factories\, and make a case for why the AI-native forge will be composable and built on open primitives.\n\nSchedule: https://ai.engin eer/code/2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1373482@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261113T013000Z DTEND:20261113T015000Z SUMMARY:Thariq Shihipar\, Independent DESCRIPTION:Speakers: Thariq Shihipar\n\nSchedule: https://ai.engineer/code /2026/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT BEGIN:VEVENT UID:code-2026-session-1380098@ai.engineer DTSTAMP:20261109T204243Z DTSTART:20261113T015000Z DTEND:20261113T020000Z SUMMARY:MC Outro DESCRIPTION:Speakers: Corey Quinn\n\nSchedule: https://ai.engineer/code/202 6/schedule LOCATION:Continental 4–6 (Mainstage)\, Hilton San Francisco Union Square\ , 333 O'Farrell St\, San Francisco\, CA 94102 URL:https://ai.engineer/code/2026/schedule CATEGORIES:talk STATUS:CONFIRMED TRANSP:OPAQUE END:VEVENT END:VCALENDAR