A reference architecture for a secure, RAG-based enterprise AI assistant built from open-source models, with policy guardrails for redaction and access control.

A reference architecture for a self-hosted, multi-tenant platform that gives every user a private agentic chatbot with permission-scoped document retrieval.

How to build multi-agent systems with the Deep Agents framework using two primitives — subagents for context isolation and skills for progressive disclosure.

Open Coding Agents and SERA make it practical for any team to fine-tune a capable coding agent on their own repository.

A reference on context engineering for AI agents — curating the model’s context window for reliability — organized as a set of composable skills covering foundations, architecture, and operations.