Desktop Agents & Pipelines

Production-grade AI retrieval systems and desktop agents that sit on top of your own data. These aren’t “toy chatbots” – they’re the backbone of serious AI workflows.

Technical Specifications

Technical Note #1: The Context Mesh Lite architecture combines hybrid vector search (dense embeddings + indexed full text), SQL, and a knowledge graph into one composite retrieval system. This ensures higher accuracy than standard RAG approaches.

RAG.PC – Local Hybrid RAG Agent

In development

RAG.PC is a desktop agent that lets you bring all your data to one place, run a Context Mesh–style RAG pipeline over it, and surface answers through a desktop UI or your favourite communication channels.

Ingest from files, documents, audio, video, and other data sources
Parse + normalize into LLM-friendly formats (Markdown / CSV with rich metadata)
Upsert into Postgres tables using a real knowledge graph powered by Apache AGE
Connect to an LLM and access via desktop UI or Slack / Discord / Telegram / WhatsApp, etc.

Stack: Postgres + Apache AGE · Hybrid RAG · Desktop UI · Multi-channel integrations

Join the RAG.PC waitlist

Context Mesh Lite – Composite Retrieval Pipeline

Context Mesh Lite is a full hybrid RAG pipeline built for Supabase to serve teams that care about retrieval quality. It utilizes a proprietary composite retrieval system architecture (see Technical Note #1 below).

You get a complete pipeline:

One main SQL query that orchestrates retrieval
Two edge functions for upsert and search
Two n8n workflows to interact with those edge functions

This is not a single “agent UI” – it’s the architecture other agents plug into when you care about precise, controllable retrieval.

Stack: SQL · Hybrid search · Knowledge graph · Edge functions · n8n

Buy Scripts (Supabase) for €79