TOBECEO

Turn one raw idea into a structured company decision flow:

• research before optimism
• department debate before commitment
• board verdict before execution
• intervention-driven replanning when assumptions change

中文文档

Most AI tools give you one polished answer.

TOBECEO gives you a process:

1. Research
2. Department plans
3. Cross-department roundtable
4. Synthesis
5. Board decision with scorecard

You can chat with agents at any stage and promote important chat turns into formal interventions.

Hero Demo

• Web Console at root path
• Flask API for project, planning, timeline, diff, and chat
• CLI for quick local planning drafts
• Persistent project/task state with SQLite
• Chinese and English conversation modes
• Stage history replay and employee-level roundtable logs

• founders pressure-testing early ideas
• product teams evaluating strategic bets
• accelerators and startup programs
• builders who want structured pushback, not generic encouragement

python -m pip install -e .[dev]

tobeceo-api

Open:

• http://127.0.0.1:8000
• http://127.0.0.1:8000/health

tobeceo-plan "AI copilot for independent gyms" \
 --summary "Help gym owners automate retention and upsell workflows." \
 --constraint "Keep CAC under control" \
 --metric "Monthly retention > 90%"

curl -X POST http://127.0.0.1:8000/api/projects \
 -H "Content-Type: application/json" \
 -d '{
 "title": "AI interviewer for junior candidates",
 "summary": "Automated pre-screening with role-specific rubrics.",
 "constraints": ["Avoid biased scoring"],
 "metrics": ["Interview completion rate > 70%"],
 "language": "en-US"
 }'

curl -X POST http://127.0.0.1:8000/api/planning/generate \
 -H "Content-Type: application/json" \
 -d '{"project_id":"<PROJECT_ID>"}'

Call generate repeatedly to move through stages.

curl -X POST http://127.0.0.1:8000/api/projects/<PROJECT_ID>/chat \
 -H "Content-Type: application/json" \
 -d '{"agent":"research", "message":"What is the biggest go-to-market risk?"}'

curl "http://127.0.0.1:8000/api/projects/<PROJECT_ID>/plans/diff?from=<V1>&to=<V2>"

1. research
2. department_design
3. roundtable
4. synthesis
5. board

The scorecard and recommendation are produced at the board stage.

Screenshot 1 Screenshot 2 Screenshot 3

Optional environment variables:

• IBC_HOST default 127.0.0.1
• IBC_PORT default 8000
• IBC_DATA_DIR default .data
• IBC_LLM_API_KEY
• IBC_LLM_BASE_URL
• IBC_LLM_MODEL
• IBC_LLM_TIMEOUT_SECONDS default 45

Without LLM config, the app still runs in deterministic demo mode.

This repo includes Render deployment support via render.yaml.

• Stronger board debate mechanics
• Checkpoint-aware selective recomputation
• Multi-user auth and workspace isolation
• Managed DB option for production
• Automated quality regression evaluation

Issues and pull requests are welcome.

If you want to contribute quickly, start with:

• new domain test cases
• better intervention heuristics
• richer board scoring dimensions
• UX improvements for plan diff and timeline views

Apache-2.0