Orca

Orca helps teams strengthen project planning from ideation to execution so they can detect weak plans, clarify priorities, and stay aligned before projects fail.

Comment

Inspiration

It all started with a reframe: instead of asking "Why do some projects collapse before reaching the finish line?", we asked "How can we help teams build and strengthen their project plans by identifying critical risks and gaps at the ideation stage — so they can focus more on the actual plan, doing less and achieving more?"

The solution was inspired by platforms like Notion, NotebookLM, Discord, and agentic IDEs. During planning, teams tend to generate more activity than actual results — jumping across multiple platforms, losing focus, and time that could have been spent discussing with the team and closing the gaps that would later cause the entire project to collapse.

What It Does

Orca turns team discussions into a structured plan — your conversations build the plan. It does not build on itself; it relies on human thinking surfaced through conversation, by:

  • Extracting key decisions
  • Identifying gaps
  • Suggesting mitigations
  • Proposing a structured plan

How We Built It

The project was split into distinct phases:

  1. Project Setup
  2. Database Schema
  3. Building the API
  4. AI Engine
  5. Frontend
  6. Integration

Throughout these phases, we leveraged the specific strengths of different coding agents:

  • Claude — refining decisions and identifying gaps in our plan
  • Codex — executing end-to-end implementation
  • Gemini — building a user-friendly interface

Altogether, we built a working MVP in 7 days.

Challenges We Ran Into

  • We frequently hit usage limits of the Gemini LLM API because we are using the free tier, which powers our AI engine.
  • When AI agents hallucinated, it wasted token budget, and we had to redo the implementation with tighter processes and stricter rules.

Accomplishments We're Proud Of

  • The AI engine works as intended.
  • API endpoints are reachable — we had zero CORS errors from day one.
  • Queue logic for agent runs works as intended.
  • Interactions feel coherent, even with an incomplete feature set.
  • Clean, polished user interface.

What We Learned

Development without a clear and structured plan is your greatest enemy — and it always starts with a problem you don't yet understand.

We learned that identifying a problem we personally experienced is the key to resonating with what we're trying to build.

What's Next for Orca

We believe Orca has the potential to grow into a full-service. Our next steps:

  • Continue development from where we left off
  • Conduct close testing with peers
  • Explore funding opportunities to begin serving real users, even at a limited scale

Built With

Share this project:

Updates