Internal automation platform
An internal tool that automates repetitive workflows and uses a locally hosted LLM via Ollama to handle FAQ and information retrieval for the team.
Role
Fullstack Developer
Timeline
2026
Type
Internal Tool
Stack
Next.js, TypeScript, Ollama, Node.js
The problem
The team was spending too much time on repetitive tasks - answering the same internal questions, manually running routine processes, and digging through documentation. There was no centralized system for automation or self-service information.
What I built
A web-based internal platform with two core features: an automation engine for scheduling and triggering routine workflows, and an AI-powered assistant that runs a local LLM through Ollama. The assistant is trained on internal documentation and can answer team questions, surface relevant procedures, and reduce the back-and-forth that was eating into everyone's time.
Why local LLM
Running the model locally via Ollama was a deliberate choice. Internal company data shouldn't leave the network - no external API calls, no data flowing through third-party servers. It also means zero API costs and full control over the model, which matters for an always-on internal tool.
Interesting decisions
The trickiest part was making the LLM responses actually useful for the team's specific context. Generic answers aren't helpful when someone asks about an internal process. I built a retrieval layer that chunks and indexes internal docs, so the model grounds its responses in real company information rather than hallucinating generic advice.