Cursor AI Engineer: From Engineering Meeting to Draft Pull Request

Ideas Feature Requests
1

Feature request for product/service

– Other –

Describe the request

I’d love to propose an idea that I think could become a natural extension of Cursor.
Today, Cursor is exceptional once an engineer starts coding. However, a lot of engineering work begins in meetings, where requirements, architecture, trade-offs, and task assignments are discussed.
Current meeting assistants stop at transcripts, summaries, and action items. What if Cursor went one step further and became an active engineering participant?

Proposal: Cursor AI Engineer

During engineering meetings (Zoom, Google Meet, Microsoft Teams, etc.), Cursor could:

  1. Join as a junior AI engineer.
  2. Understand the repository and project context.
  3. Raise clarifying questions only when highly confident that something is ambiguous or inconsistent.
  4. Suggest implementation approaches based on the existing codebase and architecture.

After the meeting, Cursor could:

  1. Convert discussions into implementation plans.
  2. Analyze the repository.
  3. Generate code.
  4. Create tests.
  5. Update documentation.
  6. Open a draft Pull Request.

For teams that prefer more control, it could generate a complete Cursor prompt instead of modifying the code automatically.
The AI would also maintain long-term memory across meetings, allowing it to remind the team about previous architectural decisions, unfinished migrations, or unresolved technical debt.

Instead of:
Meeting → Notes → Human starts coding
it becomes:
Meeting → AI participates → AI plans → AI implements → Draft PR

I believe this shifts Cursor from being an AI coding assistant to an AI engineering teammate.

I’d love to hear what the Cursor team and the community think. Would this fit Cursor’s long-term vision? What technical or UX challenges do you see?

5

You can almost do this today, just the clarifying questions/feedback loop needs to be async.
Just transcribe your meetings and create a workflow that dumps it to an agent (could easily be “review/clarify” → “plan” → “implement”).

I’d probably just attach transcripts to the initiative and hand that off if it were me.