AI Engineer
← All organizations

AI models, applications, and safety research

Anthropic

Anthropic builds Claude, a family of AI models and applications for coding, agents, and professional work. Its products serve individuals, developers, researchers, and enterprises: Claude Code is its coding agent, while Claude Science combines researchers’ tools and packages with access to computing resources and produces auditable artifacts. Developers can use its API to connect agents to external tools, with tool search and programmatic calling designed to handle thousands of tools in complex workflows.

Founded in 2021 by former OpenAI employees including CEO Dario Amodei and president Daniela Amodei, Anthropic is structured as a public benefit corporation. Its mechanistic interpretability research uses dictionary learning to identify concepts represented inside models; in 2024, researchers extracted millions of features from Claude Sonnet and manipulated them to test effects on behavior. Anthropic also introduced the Model Context Protocol, an open standard connecting AI applications to external systems, and donated it to the Linux Foundation’s Agentic AI Foundation in 2025.

In May 2026, the company reported that its annualized revenue run rate exceeded $47 billion, alongside enterprise adoption across industries. That month, Anthropic raised $65 billion in Series H funding at a $965 billion post-money valuation, including $15 billion in previously committed hyperscaler investments. The financing was intended to expand computing capacity, support safety and interpretability research, and scale its products and partnerships.

www.anthropic.com ↗

23 talks

Newest first

26 speakers at AIE

Affiliations reflect their AIE appearances, not necessarily current employment.

Barry ZhangMember of Technical StaffMahesh MuragMember of Anthropic’s technical staff and a co-creator of Agent SkillsAlexander BrickenMember of Anthropic’s technical staffJoe BayleyEnterprise AI go-to-market specialistGagan BhatMember of Technical StaffIsabella Kai HeMember of Technical StaffJamie NeuwirthAnthropic’s Head of Startup SalesZack WittenThariq ShihiparClaude CodeBoris ChernyCreator and head of Claude CodeCat WuHead of Product, Claude CodeLance MartinMember of Technical StaffKatelyn LesseAnthropic’s Head of Platform EngineeringKevin BaiMember of Technical StaffMihirJohnDavid Soria ParraCreator of MCPKaran SampathApplied AI engineerAsh PrabakerMember of technical staffAndrew WilsonTheodora ChuJohn WelshSoftware engineerEugene YanMember of Technical StaffAlex AlbertChristopher LovejoyMember of Technical StaffKarina NguyenFounder of Thoughtful

Start here

  1. How We Build Effective Agents

    Start with the essay based on Barry Zhang's talk to weigh autonomy against cost, latency, and risk, and understand why execution budgets remain difficult to enforce.

    Barry ZhangAI Engineer Summit 2025

  2. The Future of MCP

    The essay based on David Soria Parra's talk explains progressive tool discovery, client responsibility for context usage, and how programmatic orchestration can reduce repeated inference.

    David Soria ParraAI Engineer Europe 2026

  3. Claude Agent SDK [Full Workshop] — Thariq Shihipar, Anthropic

    The essay based on Thariq Shihipar's workshop shows how Bash, files, and familiar data interfaces support context gathering, alongside layered permissions and sandbox restrictions.

    Thariq ShihiparAI Engineer Code 2025

  4. How to Build Agents That Run for Hours (Without Losing the Plot)

    Anthropic Applied AI engineers Ash Prabaker and Andrew Wilson explain how to build agent harnesses that sustain autonomous software work for hours.

    Ash Prabaker · Andrew WilsonAI Engineer Europe 2026

Messages from the stage

Keep agent architecture simple and expertise reusable

Barry Zhang recommends using simpler workflows when they suffice and inspecting decisions from the agent's perspective. With Mahesh Murag, he presents skills as portable procedural knowledge layered onto a common runtime.

Separate connectivity from enterprise control

David Soria Parra describes MCP alongside skills and CLI or computer-use capabilities. Karan Sampath and John Welsh examine shared gateways for authentication, credentials, permissions, and organizational visibility.

Make verification part of sustained autonomous work

Ash Prabaker and Andrew Wilson examine persistent artifacts, context compaction, and generator-evaluator workflows. Eugene Yan applies verification to security work through reproducible exploit testing, triage, and patch feedback.

Affiliations reflect each recorded session, not necessarily current employment.

Company sources · checked 2026-08-27