One Platform. Specialized Copilots.¶
AgentOS is a composable AI agent platform — a production-grade engine with plug-and-play packs that can be configured for any company role, any industry, any workflow. Two flagship copilots are ready today. The architecture supports infinite more.
🚀 Flagship Copilots¶
Two production-grade AI copilots — each built from the same composable blocks, each a world-class specialist.
🎩 Jean-Pierre — The PM¶
- 10+ hours saved per week per PM
- Proactive risk detection — catches problems before meetings
- 17 specialized features — Sprint Forge, Delivery Pulse, Fleet View, and more
- Automated standups, reports, and sprint plans — one-click
Binary: agentos-pm · Status: ✅ Available
📊 Michelle — Analytics Intelligence¶
- 15+ hours saved per week per analytics team
- Zero hallucinations — every answer backed by actual SQL and source data
- Shared team brain — distributed learning, one teaches all
- Self-healing knowledge — learns from every correction, never repeats mistakes
Binary: agentos-michelle · Status: ✅ Available
🧱 Built Like LEGO¶
This is the real power of AgentOS: every layer is modular. The backend engine, the tool integrations, the dashboard widgets, the AI persona — they're all independent, composable blocks. Swap a pack, change a persona, plug in different tools — and you have an entirely new AI copilot for a completely different role.
Core Engine¶
A battle-tested Go backend with streaming AI orchestration, tool execution, adaptive memory, sessions, and a REST API. Shared across all flavours — proven, scalable, rock-solid.
Tool Packs¶
Modular integrations (GitHub, Jira, Slack, databases, CRMs, custom APIs) that plug into the engine. Add or remove tools without touching the core — like swapping LEGO bricks.
Frontend Components¶
A library of premium UI widgets — Kanban boards, Gantt charts, KPI cards, heatmaps, SQL Studios — that assemble into role-specific dashboards automatically.
Agent Persona¶
The AI personality, system prompt, communication style, and domain expertise. This is what makes each flavour a specialized expert — not a generic chatbot.
🔮 How Flavours Work¶
A flavour is the combination of these LEGO blocks into a tailored experience for a specific role. Each flavour defines:
| Layer | What It Configures | Jean-Pierre (PM) | Michelle (Analytics) |
|---|---|---|---|
| Agent Persona | Name, expertise, style | French PM expert, formal | Data analyst, precise, source-citing |
| Tool Pack | Which integrations | GitHub, Jira, Slack | Database connectors, SQL engine |
| Dashboard Layout | Which widgets appear | 24-card bento grid, Risk Radar | KPI engine, schema browser, SQL studio |
| Quick Actions | One-click workflows | Standup Report, Sprint Plan | Revenue Analysis, Customer Churn |
| Recipes | Automation templates | 18 PM templates | Analytics pipelines |
| Memory | Domain knowledge | Team dynamics, sprint preferences | Metric definitions, business rules |
Same engine. Same UI framework. Different configuration. That's why we can launch a new copilot for a completely different industry in days, not months.
🌐 More Copilots — Ready Today¶
Beyond the flagships, four additional flavours are available:
Yvette — Freelancer PM¶
Freelance project management: time tracking, budget management, invoice generation, daily standups, and decision logging. Built for independent consultants and developers.
Binary: agentos-freelancer
Édith — Sales Intelligence¶
CRM analytics, deal scoring, pipeline intelligence, prospect research, and automated sales reporting.
Binary: agentos-edith
Retail Ops¶
Inventory, orders, shift management, and retail analytics. Different tools, different dashboard, same powerful engine.
Binary: agentos-retail
Office Assistant¶
Document management, scheduling, and workflow automation for office productivity.
Binary: agentos-office
🌍 The Possibilities Are Endless¶
Because AgentOS is modular, new flavours can be created for any role that involves data, tools, and decisions:
Customer Success¶
Connect CRM, support tickets, and NPS data. AI-powered churn prediction, account health scoring, and automated QBR prep.
Compliance & Audit¶
Regulation tracking, policy gap analysis, audit preparation, and real-time compliance monitoring across systems.
Healthcare Admin¶
Patient flow optimization, resource allocation, scheduling intelligence, and operational KPIs for hospital management.
Logistics & Supply Chain¶
Route optimization, inventory forecasting, supplier risk analysis, and cross-dock scheduling intelligence.
Enterprise Customization
Need a custom flavour for your organization? We build tailored AI copilots that connect to your proprietary tools and data sources — using the same proven engine that powers Jean-Pierre and Michelle. Contact us to discuss your needs.
⚡ Installing a Flavour¶
Each flavour is installed with the same one-line command — just specify which one:
# Jean-Pierre PM (default)
curl -fsSL https://unicolab.github.io/agentos/install.sh | sh
# Michelle Analytics
curl -fsSL https://unicolab.github.io/agentos/install.sh | sh -s -- --flavour michelle
# Yvette Freelancer PM
curl -fsSL https://unicolab.github.io/agentos/install.sh | sh -s -- --flavour freelancer
# Édith Sales Intelligence
curl -fsSL https://unicolab.github.io/agentos/install.sh | sh -s -- --flavour edith
# Retail Ops
curl -fsSL https://unicolab.github.io/agentos/install.sh | sh -s -- --flavour retail
# Office Assistant
curl -fsSL https://unicolab.github.io/agentos/install.sh | sh -s -- --flavour office
Binary is always called agentos
The installer automatically renames the per-flavour binary to agentos (or agentos.exe). All commands work the same regardless of flavour: agentos serve, agentos chat, agentos setup.
🖱️ Double-Click Launchers¶
Every download includes launcher scripts so you can start AgentOS without opening a terminal:
| Platform | File | Action |
|---|---|---|
| macOS | Start AgentOS.command |
Double-click in Finder — clears Gatekeeper, opens browser |
| Windows | Start AgentOS.bat |
Double-click in Explorer — opens browser |
| Linux | start-agentos.sh |
Make executable, double-click in file manager |