ATOM Platform
The governed agent platform — autonomy, earned.
88% of AI agent pilots never reach production.* Atom is built for the other path.
*Industry figure (Turion 2026). Atom makes no claims about its own deployments.
What is Atom?
Atom is an open-source, self-hosted AI agent workforce — a team of specialty agents (sales, support, finance, engineering) that your people delegate to in plain language. Where other platforms sell agent capability, Atom sells agent accountability: autonomy that is earned through verified outcomes, executed inside a deterministic safety net, on your hardware.
Agents that earn trust, not assume it. Atom's agents don't just respond to commands — they operate autonomously within governed boundaries, handling routine work end-to-end. Every agent starts as a supervised intern and graduates through a 4-tier maturity model (STUDENT → INTERN → SUPERVISED → AUTONOMOUS) only after verified successful runs — 10/25/50 episodes, outcome-checked, not self-reported.
Verified outcomes, not self-report. Every mutating action is re-derived against your system of record by an independent postcondition oracle (on by default), and confidence is split into self-reported vs externally verified. An agent that says "done" is checked, not believed. A prompt-injected agent at any tier acts at that tier's scoped blast radius — bounded by a default-on sandbox layer: filesystem scope, tool whitelist, tripwires, resource caps, kill-run, egress allowlist, full provenance audit. 0.027ms P99 per check.
Your data stays yours. Workflow data, agent state, and memory live on your infrastructure — embedded store, no cloud required. LLM inference uses your own API keys (BYOK, encrypted at rest) — or local models (Ollama first-class, or any local OpenAI-compatible server: LM Studio, vLLM, llama.cpp server) for fully private deployments. EU AI Act data-governance obligations (Aug 2026)? Designed for, not retrofitted.
Free edition, full features (AGPL v3): everything in this repository — every agent, integration, and governance feature — is free and open source. Keys you configure in .env are treated as BYOK and are never gated by plans or tiers. Commercial/managed editions run this same code on the client's own infrastructure; there is no closed-source "pro" build.
💰 Budget-friendly AI agents: OpenCode Go subscription (~90% savings vs pay-per-token) — one $10/mo key unlocks general-purpose models (DeepSeek V4, Kimi K3, GLM 5.2, MiniMax M3, Qwen 3.7, Nemotron 3 Ultra, Grok 4.5) with full tool-calling & structured output — not just for coding, works for any agent workload. Setup guide →
No lock-in: 16+ LLM providers (OpenAI, Anthropic, DeepSeek, Gemini, MiniMax, Groq…) with automatic cost-aware routing, fallback, and self-healing — every run makes the next run cheaper (learning router + caching tiers).
The vision: every employee gets a personal agent team that knows their workflows, remembers context across weeks, and autonomously handles the repetitive work — research, data entry, drafting, reconciliation — while governance keeps you in control. Not AI replacing humans: AI handling the work humans shouldn't be doing.
Receipts: 0.027ms P99 governance checks (repo benchmark) · 616k ops/s cached throughput · 69+ documented TDD hardening rounds (~1,100 fixes in the deep security sweep alone) · 85k+ test functions (84,737 across 2,759 files, verified Aug 2026). External stats sourced in docs/marketing/RESEARCH_NOTES.md; copy kit in COPY_README.md + POSITIONING.md.
💻 Quick Start
Clone to first governed workflow in ~10 minutes. Start small — one workflow, one integration, one approval gate.
git clone https://github.com/rush86999/atom.git && cd atom
make setup # one-shot dev bootstrap (venv, deps, .env, frontend)
make backend # full backend on :8001
# in a second terminal:
make frontend # Next.js UI on :3001
To use LLM features, set one key in backend/.env (or add via Settings > AI):
OPENCODE_API_KEYfor low-cost subscription coding models (~90% savings, recommended) orOPENAI_API_KEY/ANTHROPIC_API_KEY/DEEPSEEK_API_KEY/GOOGLE_API_KEY… orATOM_LOCAL_ONLY=true+OLLAMA_BASE_URL=http://localhost:11434/v1for fully local
Expected result: open http://localhost:3001 → sign in as admin@example.com (password in backend/logs/bootstrap_admin_password.txt) → describe a workflow in plain language and watch it build with an approval gate before anything ships.
Full Quick Start → · Docker → · DigitalOcean 1-Click →
Solo operator? Skip setup — import a pre-built personal starter (invoice chase, candidate pipeline, support triage), each with an approval gate so nothing sends without your OK: Personal → Team Playbook →
🚀 AI-Generated Workflow Automation
Describe the outcome. Atom builds and runs the workflow.
| What you say | What Atom delivers |
|---|---|
| "When a lead comes in HubSpot, research the company, score it, create an Asana task for the rep, and ping Slack" | A governed, replayable workflow with Human-in-the-Loop approval gates |
| "Extract invoice data from Gmail PDFs, match against QuickBooks, flag discrepancies" | End-to-end pipeline: Gmail → PDF OCR → QuickBooks reconciliation → Slack alert |
| "Monitor Zendesk tickets for sentiment, auto-escalate urgent ones, draft replies" | Real-time triage agent with approval before send |
| "Generate a weekly sales report from Salesforce, format in Excel, email the team" | Scheduled workflow: SOQL query → formula-evaluated Excel → Office 365 send |
Why it's different from Zapier/Make/n8n:
- Agents, not just steps — Agents reason (not just execute): they research, decide, retry, and self-correct
- Governance built-in — Maturity gates, HITL approval, audit trail, sandbox isolation
- Self-hosted & private — Your data, your keys, your infrastructure
- Office-native — Real Excel/Word/PPTX with formula evaluation, live Canvas co-editing
- Agent-authored — You can chat with an agent to build/modify workflows (no drag-and-drop required)
Workflow Automation Guide → · Quick Start →
⚡ The AI Agent Landscape — Where Atom Fits
| 🧵 Consumer Assistants | 🧰 Developer Frameworks | 🏢 Enterprise Workforce | |
|---|---|---|---|
| Examples | ChatGPT, Claude, Notion AI, Copilot, Perplexity | LangGraph, AutoGPT, CrewAI, AutoGen | ✅ Atom |
| Interaction | Single chat, reactive only | Code-first, build-your-own | Delegate to agent teammates in plain language |
| Data & hosting | Cloud-only | Self-hosted; bring your own integrations | Self-hosted + 46+ native business integrations |
| Governance | None | DIY | Governed by design |
Atom is the only open-source platform that delivers:
- Enterprise governance (maturity tiers, HITL, audit) — without vendor lock-in
- Agents that know when to ask — autonomy earned per action type from verified track records, auto-revoked on regression; governed agent orgs with delegation contracts & privilege leases
- Self-hosted privacy — your data, your keys, your infrastructure
- Autonomous agent teammates — agents that work with your people, not just for them
- Agent-authored workflows — chat to build, no drag-and-drop
- Office/Canvas native — real Excel formulas, live co-editing
- 46+ business integrations — Salesforce, HubSpot, Slack, Jira, Stripe, QuickBooks…
📊 Comparisons
| Alternative | Focus | Key Difference | Deep Dive |
|---|---|---|---|
| Hermes Agent (Nous Research) | Personal coding/productivity assistant | Single-agent, no governance, no integrations, no sandbox | Atom vs Hermes → |
| OpenClaw | Personal productivity, messaging-first | Single-agent, Markdown memory, smart home focus | Atom vs OpenClaw → |
| LangGraph / CrewAI / AutoGen | Developer frameworks | Code-first, build-your-own governance & integrations | Why Atom? ↓ |
| Zapier / Make / n8n | Workflow automation | Step-based (not agents), no reasoning, no governance | AI-Generated Workflow Automation |
TL;DR: If you're evaluating personal agents → Hermes/OpenClaw. If you need governed multi-agent business automation → Atom.
Full feature matrix: Why Teams Choose Atom ↓
⚡ Key Capabilities
| Category | Features |
|---|---|
| 🤖 Multi-Agent Orchestration | Queen Agent (structured workflows) + Fleet Admiral (open-ended tasks) + Conductor (5 execution strategies) + validated state machine with rollback; governed fleet routing with ranked specialist matching |
| 🛡️ Governance & Safety | 4-tier maturity (Student→Autonomous), policy-gated HITL approval, comprehensive audit trail, AI-powered training, OIDC SSO + SCIM v2 provisioning + 8-role RBAC |
| ✅ Outcome Verification | Postcondition oracle re-derives success against the system of record — on by default, a refuted self-report is stamped UNVERIFIED (ATOM_ORACLE_ENFORCE kill switch); two-tier confidence provenance (self-reported vs externally verified); opt-in reviewer re-delegation loop |
| 🎚️ Agents That Know When To Ask | Every action type earns interruption-free status from its verified track record, keeps asking when evidence is thin (ask fails safe), and loses the privilege automatically when results regress — autonomy only relaxes after passing a held-out certification gate (Brier ≤ 0.25, denial-coverage ≥ 0.7) |
| 🏛️ Governed Agent Teams | Multi-agent fleets fail like human orgs — ignored instructions, redone work, nobody accountable. Atom ships the countermeasures: delegation contracts with a single accountable agent, expiring privilege leases instead of titles, conflict-of-interest detection, contribution credit feeding graduation, opt-in nightly alignment sweeps |
| 🧠 Memory & Learning | Per-turn fact extraction, 2-tier recall (SQL + LanceDB), episodic memory, memory_remember/forget, self-evolution (Memento/AlphaEvolver, self-evolving harness) |
| 🔎 Hybrid Search | documents.search fuses BM25 (FTS5/tsvector) + vector (LanceDB) via Reciprocal Rank Fusion (RRF) — semantic + precise retrieval with citations |
| 🗂️ Knowledge VFS | Agent-native document tree — ls/cat/grep/search with line-numbered citations instead of bespoke per-store queries |
| 📻 Agent Radio | Lateral peer-to-peer messaging between agents (mention-first, budget-governed) — agents coordinate without hardcoded teams |
| 💼 Office Automation | Agent-driven Excel/Word/PPTX editing on Canvas with live preview broadcast; formula-evaluating workbook runtime; agent↔document sync |
| 🧩 Mini-Apps | Agent-authored stateful canvas apps — Firecracker microVM isolation, per-instance chat |
| 🔍 GraphRAG & Intelligence | Multi-hop expansion, Leiden community detection, JIT fact verification, D3 visual explorer |
| 🌐 46+ Business Integrations | Salesforce, HubSpot, Slack, Teams, Gmail, Notion, Jira, Linear, Stripe, QuickBooks, Shopify, GitHub, GitLab, Zoom… |
| 🛰️ LLM Gateway | OpenAI/Anthropic-compatible API over your BYOK — point Claude Code, n8n, or any OpenAI-SDK app at Atom |
| 💰 Cost-Aware Routing | 5-tier cognitive classification, 16+ providers, opt-in learning router (feedback-based re-ranking), RTK token compression |
| 🤝 Interoperability | MCP client for external tool servers, ACP endpoint for standard agent clients, A2A Agent Card + message/send for agent-to-agent delegation, span tracing with optional Langfuse export |
| 🎯 Goal-Driven Loops | Agents terminate on a definition_of_done predicate instead of always burning to max_steps; utility targets, custom action surfaces, stuck-detection |
🛡️ Production-Ready Security (Default-On)
| Layer | What you get |
|---|---|
| Execution Sandbox | Filesystem scope, tool whitelist, tripwires, resource caps, KillRun — enforced at every tool-dispatch hub (in-process policy checks); mini-apps run in Firecracker microVMs |
| Encrypted Credentials | OAuth integration tokens encrypted at rest (Fernet); production fails closed without key |
| Per-Agent Capability Bindings | Zero-trust tool scoping — agent can never exceed its tier floor |
| Outbound Gatekeeper | Rate limiting, response masking, HITL mutation approval on integration calls |
| Data-Taint Tracking | Restricted data observed in a run blocks external outbound actions |
| External MCP Client | Connect to arbitrary external MCP servers (Cloudflare portals) |
Security Architecture → · Sandbox Deep-Dive →
📚 Documentation & Discoverability
🎯 Example Use Cases by Department
| Department | Scenario | Key Integrations |
|---|---|---|
| Sales | New HubSpot lead → Research company → Score → Asana task → Slack notify | HubSpot, Asana, Slack, LinkedIn |
| Finance | Gmail PDF invoice → OCR extract → QuickBooks match → Flag discrepancies | Gmail, QuickBooks, Excel, Slack |
| Support | Zendesk ticket → Sentiment analysis → Auto-escalate urgent → Draft reply | Zendesk, Slack, Email |
| HR | BambooHR new hire → Provision accounts → Invite to Slack → Schedule orientation | BambooHR, Google Workspace, Slack, Calendar |
| Engineering | GitHub PR → Run tests → Security scan → Post summary → Auto-merge if green | GitHub, GitLab, Slack, Jira |
| Marketing | Content calendar → Generate posts → Human review → Schedule multi-platform | Notion, Slack, LinkedIn, Twitter, Meta |
🏗️ Repository Layout
atom/
├── backend/ # FastAPI app — main_api_app:app (full) / minimal_app:app (smoke)
├── frontend-nextjs/ # Next.js web UI
├── mobile/ # React Native (Expo) companion app
├── menubar/ # Tauri macOS menubar companion
├── scripts/ · infra/ · installer/ · examples/
├── docs/ # project documentation
├── Dockerfile # dual-app image (backend + frontend)
└── Makefile # common tasks (start here)
🌟 Why Teams Choose Atom
Per-competitor analysis: 📊 Comparisons ↑
| Atom | Zapier/Make/n8n | LangGraph/CrewAI | OpenClaw | LangChain | |
|---|---|---|---|---|---|
| AI Agents (reason, not just execute) | ✅ | ❌ | ✅ | ✅ | ✅ |
| Governance (maturity + HITL + audit) | ✅ | ❌ | ❌ | ❌ | ❌ |
| Outcome verification (re-derived from system of record, not self-report) | ✅ | ❌ | ❌ | ❌ | ❌ |
| Default-on Sandbox (all dispatch paths) | ✅ | ❌ | ❌ | ❌ | ❌ |
| Self-Hosted / Private (your keys, your infra) | ✅ | ❌ | ✅ | ✅ | ✅ |
| Office/Canvas Native (Excel formulas, co-edit) | ✅ | ❌ | ❌ | ❌ | ❌ |
| Agent-Authored Workflows (chat to build) | ✅ | ❌ | ❌ | ❌ | ❌ |
| 46+ Business Integrations (CRM, finance, support) | ✅ | ✅ | — | 50+ personal | — |
| Cost-Aware LLM Routing (16+ providers) | ✅ | ❌ | ◐ | ◐ | ◐ |
| Mini-Apps (agent-authored stateful apps) | ✅ | ❌ | ❌ | ❌ | ❌ |
| GraphRAG / Episodic Memory | ✅ | ❌ | ◐ | ◐ | ◐ |
🤝 Contributing & Support
We welcome contributions — see CONTRIBUTING.md. Quality bar: CI-gated core suite green, typecheck clean on changed files, review required, docs updated. See docs/compliance/COMPLIANCE_MAPPING.md for the security/compliance control mapping.
- Issues: GitHub Issues
- Blog: Substack
- License: AGPL v3 — LICENSE.md
Built with FastAPI | SQLAlchemy | LangChain | Playwright | Next.js
Experience the future of self-hosted AI automation — safe enough for your whole team.
⭐ Star us on GitHub — it helps!