Waldek Mastykarz

You don't "just" replace an enterprise app

What's your minimal viable model?

Test whether your knowledge skill actually works

A bicycle for thought

What if coding agents had a learning mode?

If you built it, so can I

What happened to the paragraph?

Your product didn't get worse

Your coding agent extension needs continuous evaluation

I care less whether you wrote it

The red pen

Your app is losing its grip on your customers

Don't confuse a demo with a product

AI makes building cheap, ownership is still expensive

Does every agent skill need an eval?

What's volatile about your technology?

Your agent should start with a hypothesis, not a plan

DRY might be wrong for agent documentation

I was wrong: there are bad prompts

Your agents should learn for the organization

When organizations can build faster than they can agree

Every team needs a knowledge interface

"Coding is solved" misses the point

Inference efficiency is about protecting attention

Making room for your thoughts while reading

We're measuring AI productivity the wrong way

Every configuration option is a question we couldn't answer

The next literacy is externalizing judgment

You can't judge an AI agent until you know what the task is worth

ProxyStat is now available for Windows

ProxyStat v1.2.1 - Quick access to system proxy settings

Benchmark models using OpenAI-compatible APIs

Language model benchmarks only tell half a story

API request encoding matters

Use Prompty with Foundry Local

Integrate Microsoft 365 Copilot declarative agents with Azure AI Search

Calculate the number of language model tokens for a string

Create npm package with CommonJS and ESM support in TypeScript

Trace the location of API requests

Add background color to menu bar icon in macOS

Easily see if you have system proxy configured on macOS