Every entry is AI-native — an LLM or AI agent at its core — mapped to the department it replaces. Sales. Marketing. Support. HR. Finance. Legal. Engineering. Data. Design. Ops. Product. Executive.
📚 Docs & Tutorials · 🏭 By Industry · 🚀 Flagship apps · 🤝 Contribute
Every company is a set of departments. Every department is a set of repetitive, well-defined workflows. In 2026, almost every one of those workflows has a mature, open-source, AI-native app that can do it.
AI Workforce is the map. For each department, we curate the best open-source AI apps — LLM- or agent-native, not legacy software with an "AI" sticker — tell you honestly what each replaces, and show you how to ship it in hours.
The goal isn't to fire your team. It's to give a small team — or a solo founder — the output of a big one. Run the company you couldn't afford to staff.
- Read the Quickstart — the shared foundation (an LLM endpoint, a vector store, Docker, a domain, secrets). Local Ollama works everywhere.
- Pick your department below and see its options.
- Deploy the one-click stack that fits, or run a flagship app with zero API keys via offline demo mode.
Every listed tool is AI-native. Most departments have a specialized primary; a few (Sales, Support, HR, Finance, Ops) don't yet have a strong single-purpose AI-native tool that covers the whole job, so Veska — an open-source, AI-native ERP — is listed there as an all-in-one option alongside the individual tools.
| Department | AI-native options |
|---|---|
| 📈 Sales & Revenue | SalesGPT · OpenOutreach · our AI SDR · Veska (all-in-one) |
| 📣 Marketing & Growth | Dify · GPT Researcher · Social Media Agent · Activepieces |
| 🎧 Customer Support | Onyx · AnythingLLM · Botpress · Parlant · Veska (all-in-one) |
| 🧑💼 Human Resources | FoloUp · Aural · Hiring Agent · Veska (all-in-one) |
| 💰 Finance & Accounting | Unstract · FinRobot · Veska (all-in-one) |
| ⚖️ Legal & Compliance | AnythingLLM · Onyx · Mike · PrivateGPT |
| 🛠️ Engineering / DevOps | OpenHands · Aider · PR-Agent |
| 📊 Data & Analytics | AI Analyst · DB-GPT · Vanna · Wren AI |
| 🎨 Design & Creative | ComfyUI · InvokeAI |
| ⚙️ Operations & PM | Skyvern · Coworker · Veska (all-in-one) |
| 🔬 Product & Research | GPT Researcher · Open Deep Research |
| 🧭 Executive / Assistant | Khoj · Fabric · Meetily |
Same AI-native tools, re-sliced by vertical — 109 tools across 18 industries. Each page lists the best AI-native OSS for that sector (and flags where a vertical is still thin).
| 🏥 Healthcare | ⚖️ Legal | 💳 Fintech | 🛡️ Insurance |
| 🛒 E-commerce | 🎬 Media & Creative | 🎥 Film & Video | 🎵 Music & Audio |
| 🎓 Education | 🏠 Real Estate | 🚚 Logistics | 🍽️ Hospitality |
| 🏛️ Government | 🌾 Agriculture | 🔐 Cybersecurity | 🧪 Scientific Research |
| 🎮 Game Dev | 🤖 Robotics |
→ Full Industry Index
| Kit | Stack | You get |
|---|---|---|
| The One-Person Company | Veska | CRM + support + finance + HR + ops, run from Slack |
| The Private Brain | Ollama + AnythingLLM + Onyx | Fully offline AI RAG over your company's docs |
| The Growth Engine | Dify + Activepieces | AI content generation and campaign automation |
| The Autonomous Dev | OpenHands + Aider | Issues → pull requests, and a terminal pair-programmer |
| The Research Desk | GPT Researcher + Khoj | Cited market research + a personal AI second brain |
Original, runnable reference apps — each with an offline mode (zero API keys). Provider-neutral: point them at any OpenAI-compatible endpoint (Ollama, LiteLLM, …).
| App | What it does | Stack |
|---|---|---|
apps/ai-sdr |
Researches a lead from a URL and drafts personalized outreach | Python / FastAPI |
apps/ai-support-agent |
RAG support agent that answers from your docs with citations | Python / FastAPI |
apps/ai-analyst |
Ask your data questions in English, get answers + SQL | Python / FastAPI |
This list gets better — and more starred — with every good addition. The bar: open-source and AI-native (an LLM or AI agent at its core), self-hostable, and shippable in ~a day.
🌱 New here? Start with a good first issue. The easiest is verifying a page — re-checking that a page's entries are still accurate. It takes 20 minutes and is exactly the judgment we look for in reviewers.
Ways to contribute (each has a guided issue template): add an AI-native app · report a broken/stale entry · verify a page · propose a new department or industry · fix a deploy · or start a discussion.
How submissions are reviewed — the rubric, approval thresholds, and verification cadence — is in REVIEWING.md; the format and quality bar are in CONTRIBUTING.md. One app per PR. No affiliate links, ever.
Maintained by Arun Rajiah · Docs on hub.arunrajiah.com/docs
FAQ
What counts as "AI-native"? The tool's core is an LLM or an AI agent — it does the work with AI, not a legacy app with an AI add-on bolted on. That's why you won't find generic CRMs, ERPs, or helpdesks here.
Can I actually run a company on this? A small or solo team can run a lot of it. These are real, production-grade projects. You still need judgment, security, and (for regulated work) compliance sign-off.
Do I need API keys? No — everything is provider-agnostic and runs against a local Ollama for a fully private, free setup. Point at a hosted OpenAI-compatible endpoint if you prefer.
Are the licenses safe for commercial use? Most are MIT/Apache/AGPL/GPL. Some are source-available or have commercial gates — every card flags this; check the linked repo before you ship.