What AI can and cannot do for your business
AI is good working on small well-defined tasks within its context. To create bigger sistems where AI can bring value you need to handle things differently.
- ai
- consulting
What we learn building software and automating processes for other people. No growth-hacking, no listicles — just the things that actually changed how a business runs.
AI is good working on small well-defined tasks within its context. To create bigger sistems where AI can bring value you need to handle things differently.
Most process problems need an integration, not an application. How to tell the difference, and why the wrong one costs a great deal more than the right one.
A seven-step method for finding the repetitive, expensive work worth automating — and the cheaper option most people skip before they build anything.
A shared drive on your company network can host a real web application. How a PyInstaller executable plus a SQLite file on a file share became a deployment strategy, and when it is the wrong one.
My rule: FastAPI for APIs, Django for fullstack applications. The performance case for FastAPI, where Django rigidity costs you, and how to build APIs in Django with DRF, Django Ninja or Django Bolt.
A simple set of questions to uncover bottlenecks, repetitive work, and processes worth automating.
What I actually do as a Python backend developer, the problems I like working on, and what I expect from the people and teams I work with.
Seven real benefits of automating a business process, where each one actually comes from, and the two that usually do not survive contact with reality.
Most automation is connecting tools that already have APIs. Where the time actually goes, and how to avoid a silent failure.
A Python backend with a TypeScript frontend is the default stack for product teams. How the split works, where it pays off, and where it gets expensive.
Most projects outgrow their database choice without revisiting it. When SQLite handles real load, how far it goes, and the honest point where Postgres wins.
Scale, cash, reputation, the ability to copy you or buy you, and B2B contracts that survive a lost bid — the five advantages a large company starts every negotiation with.
Small teams move faster, they know what everyone else is working on, and they automate because it is the only way to survive. Big companies learn this ten years too late, or never.
Most automation projects fail because they automate the wrong step. A framework for deciding what to automate, what to leave alone, and what to delete entirely.
Type hints become validation and documentation, async is not an afterthought, and Pydantic keeps one definition of your data. Why I reach for FastAPI.
DuckDB is not a replacement for every database. But for analytical workloads, it can be a surprisingly better choice than SQLite or Postgres.