An open-standard agentic operating system — a unified platform that orchestrates AI agents, manages knowledge retrieval, and provides a secure enterprise brain for any organisation. Currently in production serving Carter Capner Law as CeCiL (their branded AI platform).

📊 Key Metrics

🔧 Production Components

• CeCiL.Teams — M365 Agents SDK bot for conversational RAG in Teams and Copilot
• CeCiL.Brain — Web portal with document OCR, summarisation, and medical chronology integration
• CeCiL.Indexer — CLI multi-format document ingestion, chunking, embedding, indexing
• CeCiL.VoiceRAG — Voice-based knowledge retrieval for hands-free access
• Enterprise Brain — MCP-gated vector knowledge base with hybrid BM25 + vector KNN

🏗️ Architecture

Azure AI Search with 3,072-dimensional embeddings, semantic reranking, and 1024-char chunking with 153-char overlap. Query expansion with 15 domain-specific legal synonym sets. Adaptive Card feedback loop for continuous improvement.

Deployment Options

• Cloud — Azure AI Search + Azure OpenAI (~$350/mo estimate)
• On-Premise — Ollama embeddings + local Qdrant (RTX 5090/24GB)
• Hybrid — On-prem Brain + cloud burst for peak load (recommended)

📂 Repositories

• github.com/peterjbardenhagen/AgentsOS — Platform (orchestration, brain, docs)