MCP Apps: keeping review and approval in the conversation
The agent prepares the work; the product supplies the view and controls for people to review, edit and approve it inside their assistant.
Notes from the work
Practical notes on software, teams, and AI. The decisions, tradeoffs, and questions that are worth more than a quick answer.
35 posts
The agent prepares the work; the product supplies the view and controls for people to review, edit and approve it inside their assistant.
AI on both sides of procurement can turn suggestions into requirements. Quality discovery gives vendors a meaningful project and budget to price.

I'm for sure going to post some stuff from #AllThingsOpen this week, but that got sideswiped yesterday/today for news from... PHP? That wasn't on my bingo card.
#AllThingsOpen day 2! Come stop by, second floor next to ballroom A.

I am super excited to be speaking at All Things Open Conference in Raleigh next Tuesday, October 20, about building products for people who use them through AI agents.

I've never played Go. I've played enough chess to appreciate its complexity, but Go is orders of magnitude more complex by possible positions. More than the number of atoms in the universe.
“As agents get more capable, why should your product still exist?”

Something about a Drumline always gets me pumped up. Really enjoyed this!
Conferences Hack Day, a one-day hackathon to take an idea through to a working demo. We were encouraged to build with the sponsors tools.
This gif brings me so much joy. It feels like the way I might have played a video game when I was a kid.

Last night, Dave Sifry’ gave an Ignite talk and made a good case for trying a modest model for AI code review.
I did promise to keep you posted from The AI Conference. There's a ton of topic areas here, but naturally I'm drawn to those around software development.
The AI Conference day 2 (3?)! Let’s get it.

Out here learning everything I can about AI so you don’t have to. Lots of great insights to share, stay tuned!

This is one of the most exciting things I’ve been able to be a part of in my career, and I’m so excited to see it featured.
Fine-tuned Small Language Models are the absolute backbone of an efficient AI-automation paradigm.
Anthropic quietly added what I think is a transformational capability to Cowork: a built-in browser.
I use HubSpot every day, but I have no idea how to navigate HubSpot. I’m probably not going to learn.

About a month ago, I refactored the agent network for a project I’m working on. Every agent file got a model assignment appropriate to its job, mostly Sonnet and Opus. Fable could orchestrate, but I did not want it dragging every…

Better tools are changing who can start the job, but expertise is still needed to finish the job.

Ted Lasso made this alleged Walt Whitman quote famous. "Be curious, not Judgmental."

One of the things that stuck with me from the recent All Things Open Conference meetup wasn’t one of the slides. It was the little outdoor space at The Barn: a few chairs arranged around a hearth.

Mostly agree. But sometimes the most efficient thing to do is hand this off to someone else/ their agents. Sometimes my developers get instructions about the infrastructure or devops pipelines they don’t have the context / can’t digest.…
Not every workflow needs a pit crew. Sometimes jobs are a routine process - like getting the paper tray refilled.

It’s easy to get buried in the day-to-day and forget to look up once in a while.

Spending the day at The Official Cybersecurity Summit.

Kicked off the morning at Apex Intelligence — a new Apex Chamber connection circle focused on AI for small businesses, hosted by my friend Paul Trent.
Still seeing devs post “AI can’t even count fingers / write regex” memes. That’s not skepticism. It’s denial.
OpenClaw is blowing up the internet right now — and not because it’s polished. It’s chaotic, exploitable, wide‑open, and somehow still the clearest signal of where software is going next.
Being early in AI isn’t the moat you think it is.
In a world full of new tools and louder headlines, time is our scarcest resource. Every week a new platform promises to fix everything. Most of us don’t have the space to evaluate them all, and the signal gets buried in the noise.
AI isn’t just for the tech giants anymore—it’s delivering real, measurable value for small and midsize businesses, too.
How much scale will your cloud architecture support, and how much do you need?
Let’s connect
I’m always interested in how other people are approaching the same problems.