← All organizations

Durable execution and workflow orchestration

Temporal

Temporal builds an open-source durable execution platform for developers running complex, long-lived applications. Its SDKs let teams write business logic in their preferred supported programming language, covering processes such as payments, order processing, infrastructure deployment, and AI agents. Developers run application code on their own infrastructure and connect it to either a self-hosted Temporal service or Temporal Cloud, the company’s managed commercial offering.

Founded in 2019 by CEO Samar Abbas and CTO Maxim Fateev, Temporal grew out of their work on distributed systems, including co-creating Cadence at Uber. Its technical approach separates Workflows, which coordinate business logic, from Activities, which perform external operations and can retry failures according to policy. The service preserves an execution history that replacement workers replay to reconstruct state after a crash and continue processing. This makes recovery part of the execution infrastructure rather than something each application must implement independently.

Temporal serves conventional enterprise workloads as well as AI systems: OpenAI, Replit, and Lovable use it to build agents. In 2025, the company reported more than 2,500 Temporal Cloud customers globally. It announced $300 million in Series D financing led by Andreessen Horowitz at a $5 billion valuation in February 2026.

temporal.io

Start here

  1. Building Durable, Production-Ready Agents with OpenAI SDK and Temporal

    Start with Cornelia Davis's workshop to learn how activity queues, the Temporal UI, and activity_as_tool fit into an agent implementation.

    Cornelia DavisAI Engineer Code 2025

  2. MCP Tasks (async)/ Why the heck aren't any agents supporting MCP tasks/async?

    Use the ERP invoice-processing example and FastMCP demonstrations to understand the client/server design choices behind asynchronous MCP tools.

    Cornelia DavisAI Engineer World's Fair 2026

  3. Vision: Zero Bugs

    Follow Johann Schleier-Smith's Airbus A320 and Curiosity rover examples to understand his argument for applying high-assurance engineering methods more broadly with AI.

    Johann Schleier-SmithAI Engineer Code 2025

  4. Events are the Wrong Abstraction for Your AI Agents

    Mason Egger of Temporal argues that AI agents inherit familiar distributed-systems problems and that event-driven architectures introduce fragmented APIs, queue-management overhead, ad hoc state machines, race conditions, and misleading assumptions about loose coupling.

    Mason EggerAI Engineer World's Fair 2025

Messages from the stage

Orchestration instead of scattered event handling

Mason Egger argues that event-driven agent architectures accumulate queue overhead, race conditions, and ad hoc state machines. Preeti Somal describes durable orchestration, persistent state, and operational visibility as foundations for reliable deployments.

The state behind long-running tools

Cornelia Davis demonstrates persisted LLM calls and worker recovery with the OpenAI Agents SDK and Temporal. Her MCP Tasks discussion examines how asynchronous execution adds client protocol complexity and human coordination, including a comparison of v1 and revised v2 designs.

Verification alongside durable execution

Johann Schleier-Smith presents independent verification, static analysis, defensive programming, and modular decomposition as foundations for more reliable LLM-generated software. He positions durable execution as complementary infrastructure.

Affiliations reflect each recorded session, not necessarily current employment.

Company sources · checked 2026-08-27