AI Engineer & Architect

github

Antony TC • Peerlist

<h3>About the Product</h3>
This is a production-ready AI Business OS designed to automate business processes with autonomous AI employees. It features a Graph RAG system powered by Neo4j, multi-channel communication, and supports Bring Your Own Key (BYOK) across 17 LLM providers, including custom and local endpoints via Ollama. The system includes automatic fallback mechanisms for enhanced reliability.

<h3>Key Features:</h3>
<ul>
 <li>46 Autonomous AI Employees: Capable of handling various business tasks.</li>
 <li>Graph RAG with Neo4j: Leverages graph databases for advanced retrieval augmented generation.</li>
 <li>Multi-Channel Support: Integrates with various communication channels.</li>
 <li>BYOK and Multiple LLM Providers: Flexibility in choosing and integrating with LLM services, including local Ollama endpoints.</li>
 <li>Automatic Fallback: Ensures continuous operation by switching to alternative providers when needed.</li>
 <li>Self-Hosted and MIT Licensed: Offers control and flexibility for deployment.</li>
 <li>Guardrailed Autonomy: Designed for regulated industries with safety and compliance in mind.</li>
 <li>Development Libraries: Built with 8 libraries in Python and Java.</li>
</ul>

<h3>Technical Details:</h3>
The system is developed using Python and Java, with integrations for Neo4j for graph data, Redis for task queuing, and Docker for containerization. It supports local LLM deployments through Ollama and is designed to work with various LLM providers.

<h3>Community and Reliability:</h3>
This release includes fixes for streaming answers and embedding/memory-search failures to improve reliability. It also introduces optional Redis/RQ task queues for scaling and rejection-pattern reporting for autonomy decisions. Docker Compose has been optimized for quicker startup. A significant contribution from the community has been merged for indirect graph-boosted retrieval.

<h3>Documentation:</h3>
Documentation has been updated to reflect current capabilities, including custom AI Employee role creation, Ollama-in-Docker setup, and a revised README.

RagLeap Core v0.5.0 - Reliability & Honesty

About RagLeap CoreRagLeap Core is a production-ready, self-hosted AI Business OS. Not just RAG - it's 46 autonomous AI Employees that run your entire business operations from your own infrastructure. MIT open source, no vendor lock-in.v0.4.0 - 8 Data Connectors DONE: Notion, Airtable, Razorpay, WooCommerce, Snowflake, BigQuery, Gmail OAuth2, Slack + 9 vector DBs + 46 role-scoped employees + Java port (9 modules, 87 tests).Why different?- 46 AI Roles with guardrailed autonomy: Generalist + specialized vertical roles (healthcare, legal, tax, immigration) with HIPAA/privilege compliance enforced in code, not just docs.- Role-Scoped Memory: Fixed memory leak where AI mixed roles - now 100% isolated. Your doctor AI will never see your legal data.- 8 Libraries, 2 Languages: Python + Java, Postgres + Neo4j Graph RAG + pgvector, K8s hardened, backup/DR verified.- BYOK: 17 LLM providers including Ollama local, auto fallback.- 480+ commits, 87 tests passing - built in public, production-first.Self-hosted. Regulated-industry ready. Your business logic stays yours.GitHub: github.com/antonyrag/ragleap-core