Why should your AI really be yours?

A personal AI built on open standards. Your data follows you. What it learns belongs to you. And you learn alongside other pros in your field.

Build a lab Browse labs

Your personal AI is an asset. Not an account.

In two years you'll have an AI that knows you better than your calendar. It'll know your tone, your methods, your clients, your history. Simple question: who will own it?

7-day retention

No training, no ads

Always exportable

LEGO architecture

Four technical bricks, all built on open standards. No black magic, no proprietary code. Everything readable in a text editor.

memory.md

Markdown memory

Everything your AI knows about you lives in a plain .md file. One line per fact. You can read it, edit it, version it in git.

- Tone: consultative
- ICP: SaaS B2B 50-200 emp.
- Templates: 12 emails

/skills/<name>/

Skills (Anthropic)

Each magik is a Skill folder: SKILL.md + scripts + assets. Open convention. Forkable, versionable, runnable anywhere.

/skills/
  prospect-brief/
    SKILL.md
    research.py

mcp.json

MCP integrations

Gmail, Notion, Slack, Drive… connected via Model Context Protocol. An open standard backed by Anthropic and several SaaS tools.

"mcpServers": {
  "gmail":  {...},
  "notion": {...}
}

llm.config

LLM API

Today: locked to Claude API. By design open : works with any LLM tomorrow, including local ones (Ollama, llama.cpp). Swap engines without rebuilding the lab.

provider: anthropic
model: claude-sonnet-4-6
# soon: ollama, local

data/*.csv

Database (CSV/JSON)

Structured data lives in CSV or JSON. Export at any time. Re-import in Excel, Airtable, anywhere that reads CSV.

leads.csv  (247 rows)
deals.csv  ( 18 rows)
runs.json  (1.2 MB)

Your AI lab = a folder of files

memory.md + /skills + mcp.json + llm.config + /data

It's all text. All versionable in git. All zippable and downloadable.

Do It Yourselfno vendor lock-in

Your data leaves with you.

One button. One file. Your entire personal AI system: config, magik, blocks, memory, data. Downloadable to your laptop. Re-importable elsewhere. Open format, plain readable text.

Do It Yourself export on every lab
Open documented format, no proprietary lock
Works with local AI (Ollama, Open WebUI)

Open standard: github.com/boolkeys/askmojo-spec

sales-ops-app.jsonDownload

sales-ops-app/
├── memory.md
├── mcp.json
├── llm.config
├── skills/
│   ├── prospect-brief/
│   │   ├── SKILL.md
│   │   └── research.py
│   └── weekly-pipeline/
│       └── SKILL.md
└── data/
    ├── leads.csv     (247 rows)
    ├── deals.csv     ( 18 rows)
    └── runs.json     (1.2 MB)

Your context compounds. That's what makes you irreplaceable.

Every run, your AI learns your business a bit better. After six months, it finishes your sentences.

Week 1

Onboarding

You connect your tools. Mojo learns who you are.

Month 1

Personal vocab

Your tone, your terms, your shortcuts.

Month 3

Twelve email styles

It knows when to use each one.

Month 6+

Finishes your sentences

Yours forever, never trained on others.

Learn AI with other pros in your field.

You're not alone figuring out how to use AI in your job. The community shows what others have already built. You copy, adapt, share back.

Cooperate, don't re-invent.

When one sales rep publishes their prospecting lab, twelve other reps copy and improve it. That's what learning AI as a team looks like: building domain knowledge together, without re-coding the same thing a hundred times.

Build your personal AI.
Keep it yours.

Open standards. Portable data. Domain community. Start in five minutes.

Build a lab

No credit card required.