The Unsexy Software You Already Own Might Be Your Best AI Bet
Most legacy ERPs were never bad at their job — they were just painful to use. With AI and MCP, you can wrap a modern interface around the old engine instead of replacing it.
Hugo MatthaeyRunning On-Demand GPU Inference in Europe With Terraform and Cloud Run
A walkthrough of how we set up serverless GPU infrastructure for speech-to-text and speaker diarization: what the setup looks like, what the costs are, and what didn't work on the first try.
Hugo Matthaey
91 Tools, One Server: How We Use MCP to Run Sales, Invoicing, and Inventory From a Single AI Interface
Six modules. Four backend types. Two transports. One binary. Here's how we wired our entire operations stack into a single MCP server.
Hugo Matthaey
Building Smarter Context: Search Strategies That Make AI Actually Useful
No single search strategy is good enough. Combining semantic, structured, and fuzzy search hides each method's weaknesses while stacking their strengths.
German Skrylnikov
We Built Our CRM From Scratch. Here's What 64 Versions Taught Us About Outbound
Sales tools are everywhere. Most are built by people who've never sent a cold message. We built our own — 64 releases later, here's what's inside.
Hugo Matthaey
Beyond Text: Solving Real-World Document Understanding with Vision LLMs
What if a single multimodal vision model could understand any document thrown at it? Here's what we learned building production-grade image recognition across wine labels, financial documents, and handwritten schedules.
German Skrylnikov
What 30,000 Messy Wine Entries Actually Look Like
We wrote about helping a wine ERP stop losing weekends to data cleanup. This post goes deeper — into the data itself.
Armi Piamonte
