Token spend is out of control inside many (most?) tech companies. We gathered first-hand data from 15 companies on what is happening, and what they are doing about it. The two strategies are either "let it rip" or "curb spending, NOW" newsletter.pragmaticengineer.com/p/the-pulse-...

Having projected his own humanity Claude, Richard Dawkins has fallen victim to the Eliza effect. He has unequivocally declared the mirror itself to possess consciousness when in truth he has only see a dim reflection of his self. This makes me sad.

So much of the LLM discourse now is AI-hypers trying, with all the force, to convince developers there's no need to understand how the system they own works because they can't stand the possibility that rate of generating code cannot be faster than the rate at which we're able to understand changes.

The coding has been solved. One team delivering a feature using an LLM via Jira's MCP, Skills, agents, etc. It took us 30h of rework to fix the mess 🧵 1/5

Today's post has been about how AI (LLM usage) it's teaching product teams to optimise the wrong things. emmanuelvalverderamos.substack.com/p/ai-is-teac...

This Tuesday I talk about Event Sourcing and CQRS emmanuelvalverderamos.substack.com/p/event-sour...

In the continuing saga of @ebay turning to crap, the entire site is down this morning. You'd think they'd want to get a handle on quality, but I guess AI is more important 🙄

When talking about guardrails in the context of LLM instructions, remember that these are not safeguards in an engineering or deterministic sense: they are suggestions that are subsumed into the context of a completion process.

Treat your Obsidian vault like a system. rules → scripts → skills AI as execution layer: cleaning, restructuring, evolving. Maintenance becomes cheap → so you can actually do it. www.eferro.net/2026/04/how-... Reposts appreciated if it resonates 🙏

Product strategy is not a roadmap, and a roadmap is not a timeline Too many product organisations still mistake dates for direction and sequencing for strategy. emmanuelvalverderamos.substack.com/p/product-st...

Refactoring is a liquidity strategy Why improving design is economic maintenance, not code vanity emmanuelvalverderamos.substack.com/p/refactorin...

En mayo tenemos nueva edición en abierto de nuestro curso de TDD codesai.com/curso-de-tdd/ en remoto! 🚀 Será los días 11, 12, 13, 18 y 19 de 10 a 14 (hora peninsular). 50% para particulares de colectivos poco representados. No te lo pierdas! Escribe a [email protected] 💌

Transactional Outbox Pattern, from Zero to Production 👇 emmanuelvalverderamos.substack.com/p/transactio...

Below are the 14 markers for successful Continuous Delivery within an organisation... Little bonus game... can you guess which one of these I think is the most important to get right? 👀 🧵 1/5

New article out: “Parallel Change Example: Simplifying Some Test Builders” 🚀 codesai.com/posts/2025/1... We share how a better understanding of the domain led us to simplify several test builders that were initially implemented using the Curiously Recurring Template Pattern.

I like to see how the industry has evolved. First it was scientists and engineers, then scientific studies, then the FANG companies, then influencers, and now that it’s working for me in my own business and context, it’s empirical proof that it works in the industry 🤣

Telling your teams "Use IA is not a strategy". On today's blog post I talk about why most of the companies are failling on the AI (LLM + Agents) transformation fail. emmanuelvalverderamos.substack.com/p/use-ai-is-...

1/12 Lately, I've been seeing posts like this: x.com/i/status/204... I think this framing is still too narrow. A real refactor usually adds value by default. To see why, you need to stop talking about refactoring as “cleanup” and start talking about software economics.

1/12 Most companies do not have an AI strategy. They have an AI instruction: “Use AI.” “Adopt it.” “Integrate it into your work.” That is not strategy. That is management by slogan.

Prediction: The next 12-24 months, "UX-pilled" builders will be in massive demand. Who can create intuitive interfaces, web+mobile+desktop apps that "feel good," natural, fast, and far better than the competition. THIS will be the difference vs those building "just" with AI.

Potencialmente la mejor formación de TDD del mercado. Sin lugar a dudas una experiencia muy recomendable.

The series continues! 👉 Refactoring the tests after a "Breaking Out" (peer detection techniques) codesai.com/posts/2025/1... In this post, we examine when and how to refactor the original tests.

I hate AI when it's like this: Notion used to have a one-click way to turn text into a quote. They removed this and now I have... AI So instead of sub-500ms, it now takes 10+ seconds to turn selected text into a quote (type out the prompt, then wait ~4 seconds) So backwards...

The best article going into detail about GOOS techniques 👇

The product reality in 2026

The product reality in 2026

Today's last deployment from my blog is talking about semantic diffusion emmanuelvalverderamos.substack.com/p/the-label-...

Un gran curso muy recomendable

open.substack.com/pub/bryanfin... LOVE it

I posted this thread the other day, and today I published the full article. emmanuelvalverderamos.substack.com/p/ai-assista...

Software semantic diffusion and solution explain on a diagram 🔴Problem

Gemini: Si buscas el término opuesto a "Individual Contributor" manteniendo la estructura de la frase en inglés, el término exacto es Individual Manager (aunque en el mundo corporativo se usa casi exclusivamente People Manager).

1/17 Wait, I don’t think I understand this right: “pair programming with AI”. So when I use the keyboard, am I pairing with the keyboard? Useful tool? Yes. A pair? No.

So true, thank you for sharing @booch.com

Vist a Linkedin

Nuevo en mi newsletter: ¿Qué tan bien calibrada está tu confianza? open.substack.com/pub/alfredoa...

I want LiveCodeBench but showing how many tokens/€s/KWhs it takes to get to 100%. Not interested in scores less than 100%.

Something we should all think about

From time to time, I will use an LLM for answer generation, but every time I will verify its results. I also presume that search will always be biased and inadequate, and I act accordingly. Do not surrender your humanity to tools that are architecturally and demonstrably unreliable narrators.

Please add this to your definition of refactoring, because people are still planning refactoring tasks even though this book is now in its second edition and is over 20 years old.

I caught up with Matt Klein, one of the famed developers behind the Envoy Proxy, and had a chat about AI and why the hype machine has gone too far. I think the industry is putting machines over people and the tradeoff leans too far in the rent seeker's favor.

Unpopular opinion: Most product people can’t tell the difference between a Gantt chart and a product roadmap, t’s sad, but that’s how far semantic diffusion has gone.

LLMs are like a recommendation algorithm on an e-commerce website. The problem with agentic coding is like thinking that the recommendation algorithm produces sufficiently plausible results that it's essentially you. It's there to augment your experience, not to replace it.

If you use GitHub (especially if you pay for it!!) consider doing this *immediately* Settings -> Privacy -> Disallow GitHub to train their models on your code. GitHub opted *everyone* into training. No matter if you pay for the service (like I do). WTH github.com/settings/cop...

Let's see how this goes. My new Essential Code Craft live remote training, aimed at self-funding, self-improving developers starts with workshops on Test-Driven Development on April 7 & 11. Details at www.tickettailor.com/events/codem...

Recommended

Finally a crafter meetup without AI www.meetup.com/seattle-soft...