Orkes

Orkes is the company behind the commercial Orkes Conductor platform and the main maintainer, with the community, of the open-source Conductor workflow engine. Its site (orkes.io) describes itself as a “Trusted Platform for Production AI”: “Determinism, durability, and explainability for your AI agents on the world’s most battle-tested agentic workflow platform.” Checked 2026-10-28 against orkes.io page titles and descriptions, and the Conductor OSS repository.

What it offers

• Orkes Conductor (platform) and Orkes Cloud: the managed, enterprise product. The platform page frames it as one platform for agents and workflows: business processes “span probabilistic agents and deterministic workflows”, built to run durably across both.
• Conductor OSS (conductor-oss/conductor): the open-source engine, Apache-2.0. Its README describes it as an open-source durable execution platform for microservices, AI agents and adaptive workflow graphs, which “originated at Netflix and is actively maintained by Orkes and the community”. The older Netflix/conductor repository is archived (last pushed 2023-12-22).
• Developer Playground: the pricing page says to “explore freely” in it, then move to production with enterprise-grade security, SLAs and support. No amounts are recorded here; see the pricing page.
• Use-case pages by industry (financial services, telecommunications, healthcare), Orkes Academy, a blog and events.

Conductor OSS in brief

From the README (latest release v3.33.0, 2026-10-26; repo pushed 2026-10-27; not archived):

• Durable execution. Every step is persisted; executions survive crashes and restarts, with configurable retries and timeouts, and can be inspected, restarted, rerun, paused, resumed or terminated.
• Explicit orchestration. The workflow is a versioned JSON graph; workers (any language: Java, Python, Go, JavaScript, C#, Ruby, Rust) poll for tasks and hold the business logic.
• AI agent orchestration. Native LLM tasks, MCP tool listing and calling (LIST_MCP_TOOLS, CALL_MCP_TOOL), human approval, and vector workflows for RAG. The README includes an autonomous think-act agent built from a DO_WHILE loop, an LLM chat task and a SWITCH.
• Self-hosted. Quick start is npm install -g @conductor-oss/conductor-cli then conductor server start, which needs Node.js 16+ and Java 21+; a Docker image exists. The README states 5 persistence backends and 6 message brokers.
• Its stated contrast with code-first engines: those force your code to be deterministic so the framework can replay it, while Conductor makes the engine deterministic.

How it fits

A workflow engine in the same space as Camunda (BPMN-based) and the durable-execution engines; Conductor is JSON/API-defined and aimed at microservice and AI-agent orchestration. Not to be confused with the unrelated Conductor products in the vault, such as Conductor.build and Claude Code Conductor (coding-agent tools).

Limits to know

• The pages read do not name founders or leadership, and give no funding or customer figures, so none are stated.
• Conductor is not compared here with Temporal or other durable-execution engines.

Sources

• https://orkes.io and https://orkes.io/platform, https://orkes.io/pricing
• https://github.com/conductor-oss/conductor and its README
• https://docs.conductor-oss.org
• https://netflixtechblog.com/netflix-conductor-a-microservices-orchestrator-2e8d4771bf40