RRRavish Rana
All work
Multi-AgentLive demo

Slack Agentic Bot

A persistent Slack bot that listens to every message in a workspace and autonomously structures them into tasks, content and HRBP-style 1:1 notes, all stored in PostgreSQL for later querying.

Launch live demoView source
Message classes
4
Transport
Socket Mode
Store
PostgreSQL

The problem

Valuable context, decisions, tasks and 1:1 signals scrolls past in Slack and is never captured or queryable.

How it works

A Socket Mode bot streams every message to Claude for classification and extraction, then writes structured rows to PostgreSQL so the workspace becomes a queryable knowledge base.

Highlights

  • Autonomous classification: every message becomes one of four structured types: raw message, task, content, or 1:1 HRBP note.
  • HRBP note extraction: in one-on-one channels the bot pulls issues and action items into structured records.
  • Persistent memory: all classified data lands in PostgreSQL for downstream querying and dashboards.
  • Socket Mode: no public webhook required, works behind any firewall.