All Agents Are Useless at Estimating Their Own Work
I asked four coding agents how long their own work would take, then timed them doing it. Every one was wrong, by 12× to 177×.
Michał Prządka - Blog
A personal blog site of Michał Prządka.
Latest posts
Why the gradient is a list of partial derivatives
Building the gradient formula from scratch using a ski-slope picture, with minimal calculus assumed.
Investing When Anyone Can Build (podcast highlights)
If building software is nearly free, what are VCs actually investing in?
Agents as Bounty Hunters
I built a benchmark that pits coding agents against each other in a bug-finding treasure hunt.
Hiring For Humans (podcast highlights)
What are we actually hiring for when AI can ace your interviews?
Claude Don't Code
Non-coding use cases for Claude Code and what to make of it.
This is Not Me, Dancing.
Google's video model made me dance.
Vibe Coding Works (podcast highlights)
Vibe coding has a branding problem, but if you know what you're doing, AI-assisted development can help you 10x your output.
The UX of Good Reading
Why I'm unfollowing your AI-generated content?
Cursor, Meet FAF!
I added MCP server to my productivity tool.
The Real Moat Isn't Your Agent
A battle-tested framework for evaluating LLM-based systems.
The Vibe Manifesto
A philosophy for a new generation of developers.
The Anti-Patterns of Slack
A systematic critique of Slack's design choices, revealing how its features fragment focus and productivity—and exploring smarter ways to communicate at work.
How AI Models Stack Up Against My 11-Year-Old
I created my own AI math reasoning benchmark.
Chatbots, Endpoints and Ambient Intelligence
Reflections on seamless AI integration.
My New Website
Notes on the process, costs, and lessons learned from creating my consulting business website.
Journaling
I have been journaling for 30 years.
Noa Notes - Building AI That Listens and Summarizes
Noa Notes at the AI Summit
Rocznica / Anniversary - A Love Song
A love song I wrote over 20 years ago.
Thinking with language models
Language models are not search engines and the implications of that.
The Last Decade of The Interface Worker
Reflecting on the evolving role of knowledge workers in the age of generative AI and large language models.
Gen AI Guide for a Busy Executive
An AI guide for a busy person with advice and curated resources.
LLMs in My Engineering Workflow
Exploring the impact of LLMs in my daily engineering tasks. From refining code to understanding feedback, I explain how these AI tools have enhanced efficiency and productivity in my workflow over the past 6 months.
Shazam for Singing
A small project about recognizing songs.
FAF: Automating my GTD Workflow with Langchain and GPT-4
Using Langchain to automate my GTD workflow.
Managing the Monkeys
Some thoughts after reading the HBR classic.
Prompt Engineering with Bedtime Stories
I explain how to leverage GPT's capabilities to create personalized and authentic content, using strategies like refining prompts, establishing context, and incorporating personal style.
My First Two Months As A Prompt Engineering Intern
Explore my journey with AI - from insightful moments to unexpected anecdotes, learn how artificial intelligence integrates seamlessly into everyday life.
Magic and the Stuff We Don't Understand
Some thoughts about magic, progress and LLMs.
I have a VPS!
What is my self-hosting setup and how I got there?
Hello world!
First post and hello world.