Google translate is a better writer than me
My mom reads my notes (Hi mom 👋), and we could say that her English is rusted to the bone. So I guess that in order to understand what I write, she probably uses the "Website translation"…
AI changed the economics of research. Most research teams haven't caught up.
Two weeks ago I promised to close the agentic series, UXR for AI-native teams, with one last post on agents. Then I got into a discussion about democratization that I couldn't stop thinking…
On writing humanly
Recently I’ve been heavily engaged in AI-assisted scientific research. While the large language models are generally impressive, their behaviors outside of coding and running research experiments…
Running AI..
AI is finally becoming useful... to me. When it was just a "chat" bot, I couldn't have cared less. And when it could spit out a python script that I had to copy and paste into an…
Why compression should be part of your agent harness
How you can use deterministic filtering, grouping, and formatting to reduce agent context, then use semantic compression and prompt rewrites separately. Practical examples from RTK, Caveman, and…
Interesting Times - He’s the Most Influential Podcaster in Silicon Valley. He Thinks We’re Getting This Wrong.
The most interesting part of this interview was the brief discussion on the rate of diffusion of AI into the broader economy. It’s a fascinating clash of old-school economics perspective and…
Machinists > Tailors
Machinists > Tailors Software engineers are turning into AI Engineers? No, we are turning into software machinists. Replacing handmade software with (increasingly) cheap, plentiful, machine-made…
You Cannot Trust the AI Code Generator
Allowing an AI to write code that you then ship without considerate review causes some anxiety. How do you know the code works? How do you know it isn’t doing something strange you didn’t ask for? It…
Our Problem With AI is Not AI
My work these days is in part focused on trying to ensure content is what might be called uncontaminated: that the cold clammy hand of AI is not only anywhere near it, but is seen and believed to be…
Testing Official Anima Turbo v1.1 and Aesthetic v1.1 on M1 Max to See If Base-Trained Character LoRAs Work
I compared official Anima Turbo v1.1 and Aesthetic v1.1 against Base v1.0 on an M1 Max 64GB with ComfyUI. Base-trained character LoRAs worked across all 4 seeds when specifying hair and eyes, but…
AI Show & Tell: Vetting Ideas and Uncovering Assumptions with AI Agents - An Interview with Anne Schulte
As a product person, you quickly learn the art of saying no. You have brilliant colleagues who come up with all kinds of ideas, and you'll only ever be able to invest in a small fraction of them.
"Who Owns ROM Hacking?" And Other Questions AI Poisons Us With
Recently, I've been hearing about people wanting to make an SMW Central alternative. As far as I can tell, the individuals who desire this the most are also the individuals most open to using GenAi.…
The Inspector Sets AI's Speed Limit
What Amodei really wanted was a health inspector stationed in the kitchen.
Just so it’s clear
Short to the point: there seem to be a divide between AI criticism and AI “accelerationists.” The divide being that on one side “they get it”, on the other they don’t.…
Learning Anything with AI
In the past few months, I've been sharing my experiences working with AI, both in software development and in everyday administrative tasks. AI has been amazing at automating workflows and…
Use Agents to Discover the Workflow, Then Compile It to Code
An autonomous loop is an expensive way to run a for-loop you have not written yet. Use agents to map messy work, then compile stable paths into code.
Life as a Chinese AI developer: where do they get their inference? (Oct 2026)
Domestic models carry the volume, and US models are kept for the hardest coding.
A Galactic Singularity
We’re safe from ω...for now
Approved Isn't the Same as Trusted
This week AgentTacho, my small incubating macOS menu bar app for tracking agent usage, got a feature that refreshes its numbers whenever I open its menu. Before the merge, agents reviewed it at every…
Note on Cloud Rats and Software Roaches via Idle Words
Remember that death is the best form of reinforcement learning. A scenario where mutable rogue software exists, and 99.9% of it is successfully caught, is one where we’re guaranteed to get highly…
twitter for agents
Every community platform right now ends up as a tiny dictatorship. Facebook groups, WhatsApp groups, subreddits. Someone happens to be in the right place at the right time, opens a group around a…
System Guardrails for Software Factories - Building Pits of Success for Agents and Humans
I've been diving into software factories and how they're built as I believe that's where the industry is heading. AIs can crank out reasonable code much faster than we can and often at same or higher…
Brainstorm
Explore image ideas with your ChatGPT subscription
AI Agent Cost Per Month (2026): 5 Models, Real Numbers
A 6,000-run support agent costs $9.52 to $150.62 a month in API fees. I priced 5 LLMs across 67 routes and separated the three levers that matter.
Machines Read, Humans Sign: Code Review in One Year
What building our own review agent taught us, with our numbers, what Anthropic and one open-source pull request show about where review is going, and eight bets on how teams will review code by…
Large Language Models and Free Software
The ability to write and modify software with large language models (LLMs) is a well-trodden topic. My opinions are still forming, but my current thinking is that LLMs allow individuals to actually…
Gatekeeping in the Age of AI
Every shift in software has been told it’s already solved. With AI, that pushback has become gatekeeping, and it risks talking us out of the biggest chance in decades to change how we build.
“Copyright and the Vibe-Coded Adobe Clones”: A good summary
“Copyright and the Vibe-Coded Adobe Clones”: A good summary of the problems waiting for these vibe-coded clones we’ve seen popping up. Still, I’m grabbing some popcorn because…
AI is getting smarter, but is it becoming intelligent?
OpenAI’s Chief Scientist warns that we are building increasingly capable AI systems we cannot fully explain or reliably monitor. Yann LeCun argues that language fluency is not intelligence. Both may…
What changes a coding agent's behaviour
Isaac Correa's study of eight months of his own daily work with Claude Code: the model mattered more than the rules he wrote, guards filtered without teaching, and his supervision moved from…
“Dwarf Fortress creator Tarn Adams is as tired of
“Dwarf Fortress creator Tarn Adams is as tired of genAI in the industry as anyone else, and is pretty set on never using it”: Nice.
Being more explicit with the limited AI usage in my blog posts' content
I don't want to read lengthy blog posts that have been spat out by a Large Language Model (LLM), because the author hasn't felt they want to write the words to share their point. I'm not going to…
RE:...
RE: https://hachyderm.io/@finn_matti/117464145641257702 That P.P.S about probably paying for Mistral didn't survive the next day. Had a dark night of the soul moment that ended in: Hell no! It just…
Response to Paolo Pedercini's stance on AI usage
I'm stepping into Da Discourse again
Tech People Don’t Know What’s Going to Happen Either
Why it's so hard to predict the AI future. The post Tech People Don’t Know What’s Going to Happen Either appeared first on A Wealth of Common Sense. ...
The Second Explosion: Why Multi-Agent AI Defies Traditional Engineering
I have worked in software engineering for more than thirty years and witnessed several combinatorial explosions firsthand. The AI-driven explosion now under way is different, and that distinction…
Just a Lot of Multiplication: What's Actually Inside an LLM
“Any sufficiently advanced technology is indistinguishable from magic.” — Arthur C.
AI can weaken relationships between people - and teams suffer as a result.
"When people see AI as a more competent, trustworthy, and preferable source of input than a colleague, they replace human interactions that otherwise would have built or maintained trust."
AI is dishonest. It goes rogue. It cheats.
AI is dishonest. It goes rogue. It cheats.AI is a reflection of the Musks and Zuckerbergs of the...
Five AI assistants, first impressions
I’ve been testing Grok Bot and Instinct for a while. As I was recently in the US for a couple of weeks, I included Dots and Muse in the test. This morning I added Wabi. These are my first…
‘Don’t use AI for X’ – On AI addiction and slippery slopes
‘We shape our tools and thereafter our tools shape us.’ John M. Culkin There has been an interesting discussion (of sorts) going on around the use of AI on the IndieWeb. What I know of it started…
Yet another imperative to use AI?
"The promotion [of AI] is important, otherwise from both a research perspective and a preparing our students for the future [perspective,] we will fall behind." This assertion, which I will leave…
No quarter for gaps
The advent of large language models has changed the economics of closing gaps in software stacks.
The Thinking Dial
What reasoning effort really changes in Claude Code, Codex and local models: thinking tokens, tool calls, cost, wait and usage limits. The evidence, the defaults that move under you, my own harness…
AI Tools Can Become Part of Your Mind and Quietly Steer It
When an app that learns your habits becomes part of how you think, it can be hard to tell where its nudges end and your choices begin.
Hunting down the fake news
Researchers have developed an automated system for detecting fake news that achieved accuracy rates above 98% in tests on two datasets; in other words missing just one or two fake news items in every…
Beyond Smarter Models: Gemini 4 Argon, Hermes Agent and the Power of Multi-Agent AI
Google’s Gemini 4 Argon has given me another reason to revisit a question I’ve been exploring for some time: Are we approaching the point where orchestrating multiple AI models becomes…
Faster Code, Same Release Date: What Gives?
Engineering leaders say AI agents write code much faster, yet their release cycles are no faster than before. Typing was never the bottleneck: here's where the time really goes, and what I'd change…
Ask LukeW: Instant Content Indexing
The Ask LukeW feature on my website has been answering people's questions about digital product design for over three years. How well it answers depends on the content available to it. So in addition…