[achushankar]

Hi HN! I built Syllabi – an open-source platform for creating agentic AI systems that integrate tools, use knowledge bases, and deploy across channels.

The Problem: I kept needing AI that could both answer questions from company knowledge AND take actions (send Slack messages, trigger workflows, call APIs). Existing solutions either don't support agentic tool use well, lock you into their cloud, or require weeks to build from scratch.

What Syllabi Does (Three Pillars):

1. INTEGRATE ANY TOOLS • Call webhooks & custom APIs • Send Slack messages, emails, calendar events • Trigger workflows in external services • Connect YOUR custom tools via API/webhooks • AI intelligently decides WHEN and HOW to use each tool (that's the agentic part)

2. KNOWLEDGE BASE (RAG) • Transform docs, videos, websites into knowledge base • PDFs, Google Drive, Notion, Confluence • Advanced RAG with source citations • Click citations to see exact passages highlighted in original documents • Multi-format processing with smart chunking

3. DEPLOY ANYWHERE • Embed widget on any website • Slack & Discord bots • Microsoft Teams (coming soon) • Standalone web app • REST API for custom integrations • One agent, multiple channels

Key Technical Features: - MIT licensed, self-hosted, privacy-first - Modern AI models (latest GPT-4, GPT-4o, o1 series) - More providers coming (Anthropic, local models) - Agentic tool selection with function calling - Channel-agnostic core with adapter pattern - Async job queue for document processing - Plugin system for custom skills

Tech Stack: Next.js (frontend), Python FastAPI (backend), PostgreSQL, Supabase, OpenAI API

Use Cases: • AI course assistant that answers questions AND books office hours • Support bot trained on docs that can create tickets in Linear/Jira • Team knowledge base in Slack that triggers workflows • API docs helper that generates AND runs code examples

Architecture Highlights: - Modular design with clean separation of concerns - Row-level security for multi-tenancy - Docker deployment for easy self-hosting - Comprehensive API for custom integrations

I started building this 6 months ago because every project seemed to need the same thing: an AI that could access knowledge AND take actions, without vendor lock-in or per-message pricing.

Website: https://www.syllabi-ai.com/ GitHub: https://github.com/Achu-shankar/Syllabi Docs: https://www.syllabi-ai.com/docs

Would love feedback from the HN community – especially on: - Agentic AI architecture approaches - Tool use and function calling strategies - Multi-channel deployment patterns - Self-hosting and security best practices

Happy to answer questions about the tech stack, RAG implementation, agentic tool selection, or anything else!

[diffeomorphism]

The post title should probably start with "Show HN:".

What kind of security guarantees do you have?

It seems to meet that your "problem" usually is unanswered on purpose:

https://simonwillison.net/2025/Jun/16/the-lethal-trifecta/

This has access to sensitive knowledge, tool use and exfiltration. So, the tech seems nice, but I doubt I could ever get permission to deploy this.

[achushankar]

Excellent point about the security concerns. You're right that the combination of: - Sensitive knowledge access - Tool use/actions - Potential exfiltration

Is a serious concern, especially in enterprise environments.

Currently, this has: - Row-level security in Supabase - API key auth - Rate limiting

But it does NOT have: - Comprehensive audit logging - Fine-grained permission controls - Tool execution sandboxing - Data loss prevention

You're right that this shouldn't be deployed in production with sensitive data without significant security hardening. I should have been clearer about that.

Thanks for the link to Simon's article - very relevant. This is more suited for learning/experimentation than production use with sensitive data right now.

If anyone wants to work on security features, I'd be happy to collaborate on that!