Model Context Protocol (MCP): 3 Misconceptions and Fixes

I Managed a Swarm of 20 AI Agents for a Week and Built a Product. Here Are the 8 Rules I Learned. | zach wills

Query your real-time cloud cost and usage data from Claude, Cursor, and ChatGPT - DoiT Cloud Intelligence™ DoiT Cloud Intelligence™ changelog.doit.com/query-your-r...

Completely forgot to mention it here, but I've hooked up my receipt printer to a public website through a Raspberry Pi and Cloudflare. You can send me a message, and it'll print out immediately on my desk! ping.aschmelyun.com

Quality will always find a market. Even with AI.

The good folks at the edge of AI in San Francisco are discovering Requirements Engineering (which I had in college back in 2002) and why it matters: www.youtube.com/watch?v=Vvkh...

Powerful combination: the one goal meets small everyday steps

Audible’s recommendations are top notch these days 😂

Pro-tip: when interacting with GitHub using an LLM (via MCP or gh cli): tell it to avoid using “gh search”. Results will be poor and will burn through tokens.

Good managers prune.

Allen curve and how we are more connected the closer we are physically.

Clickbaity title (aren't they all these days), but with interesting thoughts on where and how agents work well or not: utkarshkanwat.com/writing/bett...

Velocity cures a lot of things. Be the support.

Hands down the best write-up on LLMs and Agents that I've read harper.blog/2025/04/17/a...

I know when we achieved super intelligence (AGI or whatever we’re calling it these days) when I see a robot properly folding my family’s clothes given all the variation they entail.

Flower time in Brussels

When disruption (e.g. LLMs) comes about slicing through everything, sometimes what is needed is a way to mend the cuts.

No poop, happy passersby

Roses and sunflowers in a jar

Expectations, Happiness, and a happy 2025! ✨

Crowded

Seth Godin, as usual, has a great point on pushback against AI. The framing is not if, it's how. seths.blog/2025/07/prod...

The illusion of competence – by making it look easy, you fool them into thinking it's not hard.

When the org is in survival mode, don’t expect the employees to be in thriving mode.

When being an EM is both great (moat from AI) and terrible (you basically have to rebuild from scratch with every new organization.

Something positive today: 30 minutes with a stranger visualization by the amazing folks at Pudding buff.ly/vc577tR

All models are wrong, but some models are useful. A meta article about how we model human systems an why the model can't be neat or clean. longform.asmartbear.com/models/

Andrej Karpathy does it again by taking the current state of the art of LLMs and explaining it to the rest of us. Recommend to watch video: www.youtube.com/watch?v=LCEm...

"When we see a lot of fires, our urge is often to build a brand-new fire station. Something that will fix and solve all underlying root cause problems. However, in reality, all we might need is to find the nearest fire hydrant to hook up to." I've been guilty of this. canopy.is/shared_train...

On How Long it Takes to Know if a Job is Right charity.wtf/2025/06/08/o... "Have you ever heard the metaphor of the chicken vs the pig? The chicken contributes an egg to breakfast, the pig contributes bacon. The punch line goes something like, “the chicken is involved, but the pig is committed!” ..

This 👇💯

Rise of the “operator EM”, how the EM role changed in the last couple of years.

“Prompts as Source Code (…) what if we treated prompts as the actual source code? Imagine version control systems where you commit the prompts used to generate features rather than the resulting implementation. When models inevitably improve, you could connect the latest version and regenerate”

New blog post of mine: Why I work at a time tracking company migueldavid.eu/posts/why-i-...

I have definitely been there. You?

Housepecker

Be(e) on the wall

Gather around children

Cogs on the wall

BE

Colors

From the amazing Will Larson, 3 simple rules on how to provide feedback on documents.

✍️ How to instantly feel better, really. readwise.io/reader/share...

TIL that GitHub Copilot (at least on the web), can't iterate on its own paging, even when a larger window is requested. #disappointing #copilot #ai

Ice cream weather

Let’s practice sacrifice