Your Company Knows a Lot. The Problem Is Nobody Can Find Anything.
I’ve worked at several IT companies. Different industries, different teams, different tech stacks. But one problem showed up everywhere.
Finding information that definitely exists — somewhere.
The Problem Nobody Tracks
Every growing company generates knowledge. Procedures, pricing, customer data, reports, onboarding guides. Over time, that knowledge stops being an asset and starts being a maze.
A company wiki? Usually there is one. But finding anything in it means guessing the exact keyword someone used when they wrote it. A slightly different phrasing — and the results are useless. A classic search engine doesn’t understand context. It matches words.
Everything else lives in scattered files across shared drives, servers, and local machines. Everyone knows it’s in there somewhere. Nobody knows exactly where.
The result? Thirty minutes hunting for a single piece of information. Several times a week. Per person.
Nobody measures this. But everyone feels it.
Growing companies feel it most acutely. New hires ask the same questions the previous cohort asked. Veterans spend their time answering instead of doing. Knowledge doesn’t scale — only the people carrying it in their heads do.
And then there’s the database problem. To answer a simple question — “how many orders did we process last quarter, broken down by region?” — someone either needs to know SQL, or needs to wait for a developer or analyst to free up time. Who have their own priorities.
The real cost of information chaos isn’t the time spent searching. It’s decisions made on incomplete data — and knowledge that walks out the door when someone leaves.
How I Tried to Solve This
In December 2024, I built the first version of an application designed to fix this problem — at least for myself.
The goal was simple: one place with fast, easy access to all the information I needed. No keyword guessing. No digging through dozens of files. Just a plain-language question and an answer in seconds.
The technology that makes this possible is called RAG — Retrieval-Augmented Generation. I wrote about it here. In plain terms: instead of answering from its own training data, the AI model first searches a defined document base, finds the relevant passage, and builds its answer from that. Every response has a source. The system doesn’t guess — it quotes.
The first prototype gave me my own private “ChatGPT” running on my local machine, powered entirely by my own data. Everything stored in my own database, the AI model under my control.
It works. And it’s fast.
What the System Can Actually Do
Over time I expanded the application to handle more data sources. Today it supports three:
Local files — PDFs, Word documents, Excel spreadsheets. Load them once, query them as many times as you like. Useful for procedures, SOPs, proposals, contracts.
Web pages and articles — paste a URL, the system pulls the content and adds it to the knowledge base. Useful for teams tracking technical documentation, supplier materials, or industry publications.
SQL databases — and this is where things get interesting for entire departments. The user asks a question in plain English. The system generates the SQL query, pulls the data, and returns a readable answer.
Non-technical team members can now get answers from the database independently — no developer, no analyst, no waiting.
What This Means for Data Security
This question always comes up — and rightly so.
The system can run entirely on-premise. The language model runs locally via Ollama, data stays in the company’s own database, and nothing gets sent to external servers.
For businesses handling sensitive data — financial records, HR information, commercial contracts — this isn’t a nice-to-have. It’s a hard requirement.
For businesses without those constraints, the system also supports API integration with external model providers: OpenAI, Anthropic, and others. The choice stays with the client.
Who Gets the Most Out of This
Not every company feels this problem with the same intensity. In my experience, the value is highest where:
- knowledge is scattered across multiple files and systems, and finding it regularly eats up significant time
- the company is actively hiring and onboarding consumes experienced team members’ time
- data lives in a SQL database, but accessing it requires a developer or analyst in the loop
- working with sensitive data rules out sending anything to external AI tools
One Thought to Close With
Your company already has the knowledge. The problem is it’s everywhere — which means, in practice, it’s nowhere.
RAG doesn’t create new knowledge. It makes the knowledge you already have visible — available to every team member, at any time, without searching and without going through someone else first.
I build these systems to order. If anything in this article sounds like your day-to-day — I’d be glad to talk through what it might look like for your business specifically.