AI-Powered Knowledge Base Pipeline Development
Michael Limisi
System 04 — Case Study
An AI-powered pipeline where engineers paste a chat log into Claude, which automatically extracts the core problem, root cause, and resolution steps, then publishes a structured incident report directly to a Notion Knowledge Base via Make.com automation — in seconds, with no manual writing.
Problem
A team's hard-won knowledge ends up scattered across chat logs, closed tickets, and stray docs — and none of it is findable when someone hits the same problem again. Issues get re-solved from scratch because nothing became reusable.
What I Built
A pipeline that turns raw, messy sources into a structured, searchable knowledge base automatically — taking chat logs and documents, using AI to summarize and organize them into consistent entries, and filing them where the team will actually look.
Architecture / How It Works
Source content flows through an AI structuring step that summarizes and standardizes it, then lands as clean entries in a knowledge base such as Notion. Designed to run continuously, so new material keeps flowing in rather than requiring a manual cleanup each time.
01Claude (MCP Client)
02Python (Knowledge Engine)
03Make.com (Orchestration)
04Notion(Structured KB)
Outcome
Throwaway conversations become institutional memory — problems get solved once and stay solved, instead of being re-debugged every time they resurface.
Screenshots
Autonomous Knowledge Builder screenshot 1
Autonomous Knowledge Builder screenshot 2
Autonomous Knowledge Builder screenshot 3
Let's fix what's actually broken.
I'm actively looking for remote AI engineering and operations roles where I can ship real automation, not just talk about it.
Built for impact, not just looks. · Michael Alusa Limisi · Nairobi, Kenya · © 2026
Like this project
Posted Jul 30, 2026
An automated pipeline transforms messy chat logs and documents into a structured, searchable knowledge base, summarizing and organizing them for easy access.
