Julep review
A Python framework for graph-based agent workflows, with optional durable execution and self-hosting.
Reviewed by iTechGuides Editors · Editorial team · Updated Oct 2026
Julep is an open-source Python framework for building AI-agent workflows, aimed at developers who want to author flows as code and control how tools and reasoners interact. Its @flow model compiles workflows into a graph intermediate representation, where developers can combine registered tools, pure functions and LLM reasoners. It is a fit for teams seeking a self-hosted option and workflow-level control; it is less suited to buyers looking for a managed, ready-to-use application platform.
The workflow model is its central strength. Branches, fan-out, retries, timeouts and rescheduling support more involved execution paths, while local deterministic dry runs can help inspect behavior before deployment. For durable execution, Julep offers optional Temporal and DBOS support rather than making either a prerequisite. Its CLI covers listing, graphing, running, testing and deploying agents. Tool options include user-defined, system, integration and API-call tools, as well as MCP references with schema snapshots. Document embeddings, full-text search, vector search and hybrid search broaden its scope to retrieval-backed workflows.
The integration footprint named for Julep centers on Temporal, DBOS, MCP, OpenTelemetry and Langfuse. That is a focused ecosystem rather than a broad catalog of application connections, so teams should check whether their required services fit these tools or can be connected through custom tools. The project also distinguishes Julep 3, described as a release candidate, from its preserved v1 agents API; teams should identify which interface their implementation depends on. Choose Julep when Python-authored flows, tool control, self-hosting and optional durable execution matter most. Consider another framework if a stable, unified API or a wider prebuilt integration ecosystem is a primary requirement.
Julep pros and cons
- Where it wins
- Compiles Python @flow workflows into a graph intermediate representation
- Supports branches, retries, timeouts, rescheduling and deterministic local dry runs
- Combines tool references, agent tools and hybrid document search
- Where it doesn't
- Durable execution depends on optional Temporal or DBOS
- The documented integration set is focused on workflow and observability tools
- The project distinguishes its release-candidate Julep 3 from the preserved v1 agents API
Julep fact sheet, pricing and score →
Advertiser disclosure: iTechGuides is reader-supported. We may earn a commission when you click some links. How we rank.
Last updated · How we research and update