LangGraph Project Planning Assistant

The problem

A short project idea can leave important requirements unstated. The goal was to turn that brief into a structured plan a developer could inspect, rather than hide assumptions inside one long AI answer.

What I built

• A clarification stage that asks follow-up questions when the brief is incomplete.

• A supervisor that assigns research topics to specialist research agents.

• Research-tool execution and a compression stage that summarizes findings for the supervisor.

• A report-generation stage that combines the research brief and findings into a structured Markdown plan.

• A human-review step for protected file-tool calls.

• Modular Python code, model and research configuration, tests, and a repository example plan.

Inputs and handoff

The input is a project description, refined through clarification. The report prompt organizes the plan around technology choices, features, architecture and implementation phases. Configuration controls the research tools, models and iteration limits. The public repository includes an example Markdown plan; the gallery's short CSV-import plan is a separate fictional illustration.

The outcome

The implementation makes the path from requirements to research and a plan inspectable. A developer can use the draft to discuss scope and dependencies before implementation. Technical review remains necessary for the proposed design, estimates and assumptions.

Who this helps

Product and engineering teams exploring an assistant for preparing project plans. This work illustrates capabilities relevant to an AI Workflow Buildout.

For a similar project, share a short brief, the plan format you need and the decisions that should stay with your team.