A guided AI skill that walks you through building a complete AI-powered organization on Paperclip — from your first idea to a ready-to-import company.
No blank forms. No prior experience needed. Just a conversation.
Paperclip Org Builder is a skill for AI assistants (like GitHub Copilot) that turns a brainstorming session into a fully configured AI company you can import into Paperclip.
You describe what you want to build. The AI proposes the whole thing — team structure, agent roles, schedules, goals, budgets. You approve, adjust, and move on. At the end, you get a package you can import directly.
The skill guides you through 9 stages, one conversation at a time. Your progress is saved after every stage — you can stop and come back whenever you want.
Stage 1 → Vision What's your company for?
Stage 2 → Identity Name, brand, conventions
Stage 3 → Org Chart Who works here and how do they connect?
Stage 4 → Agents Each agent's full operating instructions
Stage 5 → Skills What tools and skills should agents have?
Stage 6 → Goals What is the company trying to achieve?
Stage 7 → Budget How much does each agent cost per month?
Stage 8 → Routines What recurring tasks run automatically?
Stage 9 → Export Review and download your company package
Every stage follows the same pattern:
- The AI reads what you've built so far
- It proposes a complete draft for this stage
- You review, adjust anything you want, and say "move on"
- Done — next stage loads automatically
Just tell your AI assistant to set up a company. Any of these work:
- "I want to create an organization"
- "Help me build a company on Paperclip"
- "Set up my org"
The skill activates and starts the conversation.
A complete, ready-to-use AI company — every team member configured with their own role, responsibilities, and working style. Import it into Paperclip and your team starts working immediately.
At the end of the process, the AI gives you a single command to run:
npx paperclipai company import ./your-companyyour-company/
├── Company profile # Identity, mission, brand
├── Team configuration # Every agent: who they are, what they do, how they work
├── Project list # Goals and active projects
└── Schedule # Recurring tasks that run automatically
Every person on your AI team gets their own profile — drafted by the AI, reviewed by you.
| Part | What It Defines |
|---|---|
| Identity | Job title, responsibilities, what they will and won't do |
| Personality | Values, quality standards, how they handle edge cases |
| Working style | A step-by-step checklist they follow every time they work |
| Access | Which tools and resources they can use |
You're always in control. At every stage, the AI shows you a proposal and asks:
"Anything to adjust, or shall we move on?"
You can:
- Say "looks good" to approve and advance
- Edit specific parts — the AI incorporates your changes
- Go back to an earlier stage if you change your mind (e.g., rethinking your team after seeing the budget in Stage 7)
For larger teams, the AI groups reviews to avoid fatigue:
| Who | How |
|---|---|
| CEO | Reviewed individually, in full detail |
| Managers | Reviewed as a group |
| Workers | First of each role reviewed, then "Apply this to all 3 remaining engineers?" |
A counter keeps you oriented throughout: Agent 3 of 10 · Workers — Engineers · 2 approved
Everything is saved automatically after each stage. If you close the chat or need to take a break, just start again — the AI will pick up exactly where you left off, even if you were mid-stage.
You can revisit any earlier stage at any time. Just say it:
"Actually, I want to change the org chart" "Let's revisit the budget"
The AI navigates back, shows you what's there, lets you change it, and re-advances through anything affected downstream.
The skill quietly checks your work at every stage — catching issues as they happen rather than surfacing them all at the end.
| Stage | What Gets Checked |
|---|---|
| After naming your company | Name format, unique identifier |
| After building your org chart | Exactly one CEO, everyone has a manager |
| After configuring agents | All team members have instructions |
| After setting goals | Every goal and project has an owner who exists |
| After setting budgets | Numbers add up, nothing is negative |
| Before export | Full review of the entire package |
If something looks off, you're told right away — not after hours of work.
MIT