Skip to content

Home

AEGIS — Adaptive Execution Graph Intelligence System
Status License: AGPL-3.0 Roadmap Python 3.11+ Last Commit

AEGIS is not a chatbot. It is not a wrapper around an LLM.

It is a control plane — an operating system layer that sits above AI runtimes and governs them. Where most AI projects ask "which model should I call?", AEGIS asks "what capability is required?"

What makes AEGIS different

Most AI systems are built around a specific model or runtime. Swap the model, and the system breaks. AEGIS inverts this: every AI engine — Anthropic, OpenAI, Ollama, a local model — is just an adapter behind a fixed interface. AEGIS decides what capability is needed, finds the best available engine that can deliver it, and routes the work there. No engine is load-bearing.

### Control Plane AEGIS wraps AI runtimes externally — governing them without coupling to any single one. As AEGIS matures, it retires its runtime dependencies one capability at a time.
### Cognitive Fabric Every AI engine sits behind a fixed interface with a single capability vocabulary: `think`, `plan`, `embed`, `summarize`, `vision`, `stream`. Adapters translate — they never reason.
### DAG Engine All non-trivial execution is a directed acyclic graph — deterministic, persistent, and resumable. An interrupted task picks up exactly where it left off.
### Module System New capabilities arrive as installable modules. No capability bloats the core. Departments, integrations, and add-ins are built from standardized templates and are fully uninstallable.

Current status

AEGIS is at v1.1.0 and entering Phase 12 — Public Website + Distribution. Phases 0–11 are complete and tagged — through the module marketplace, production hardening, native desktop app (Tauri), Android app (Capacitor), Whisper STT + Piper TTS voice with wake word, and Memory Fabric tags.

Phase Name Status
0 Governance & Repo Foundation ✅ Complete
1 Runtime Baseline ✅ Complete
2 Control Plane Core ✅ Complete
3 Cognitive Layer ✅ Complete
4 Execution Graph (DAG Engine) ✅ Complete
5 Agent System ✅ Complete
6 Shared Services + API ✅ Complete
7 Module System + AI Skills ✅ Complete
8 UI (Web App Foundation) ✅ Complete
9 Ecosystem / Marketplace ✅ Complete
10 Production Hardening ✅ Complete
11 Desktop + Mobile Apps ✅ Complete
12 Public Website + Distribution 🔄 Next
13 Knowledge Foundation ⏳ Planned
14 Obsidian Bridge ⏳ Planned
15 Platform Polish ⏳ Planned
16 Connected Services ⏳ Planned
17 Background Intelligence ⏳ Planned
18 Workspace & Projects ⏳ Planned

See the full Roadmap → for scope, exit criteria, and the Runtime Independence Thread.

Get started

Governance-first · Memory-native · Runtime-agnostic