@simonwillison.net hi 👋 I built a way for Claude Code to avoid the Lethal Trifecta through tool call ordering and plan mode, building on your work and on CaMeL. Means far fewer approvals and easier to parse statements of intent. Grateful for feedback! github.com/chrismdp/loc...

@chrismdp.com post is well worth a read. Rather squashes the idea that AI will give us more than incremental productivity gains. Related, I'm seeing a trend back to BDUF with GenAI. For complex work, we just don’t know everything upfront. Is why agile/iterative approaches came about

FINAL CALL: "Kill Your Prompts" starts in one hour! Free to join? Sign up: chrismdp.com/webinar All sign ups get full slides and recording (even if you don't make it)

You control less of your AI conversation than you think. (Quite proud of this infographic. Didn't do well on LinkedIn. Perhaps BlueSky users are more discerning?)

I have recently gone back to Notion after a long break. Why? MCP. Changes everything.

New website (eek): what do you think? www.chrismdp.com

Bitten the bullet and paid for Opus and the Claude Max plan for Claude Code. Jury is out so far on whether it is worth it: trying to be objective on usage and state hypotheses up front.

you see this kind of reaction is why I use Claude over ChatGPT 😁 (although I'm already missing the per project memory)

Whew: more LinkedIn posts scheduled for the next 3 days. It's exhausting coming up with actually good AI content all the time!

Just tried local MCP servers, again. Bleurgh. I'm pretty technical and it took an hour to get one of the most popular ones working at all, and then the results were extremely poor. MCP local is a rabbit hole. Don't bother. Remote, well maintained MCP servers is the way to go.

Brilliant day at London Tech Week and Founder Gathr event. Super tired now, so writeup will have to wait until tomorrow (watch out for it on LinkedIn). Some amazing conversations!

Can I have sub-projects in my AI tools, with layers of context please? I want a folder structure, or the ability to write a prompt in multiple projects at once to get all the context

The thing that sets apart @cursor.com.web.brid.gy from over AI coding IDEs for me is not the agents/prompting: it's the UX. The tiny friction reducers like amazing smart autocomplete, tools in Inline Assist, easy checkpointing, automatic linting fixes, etc.

Just been trying Zed with Gemini 2.5 for coding for a morning. Takeaways:

Finding more and more I'm just using the main Desktop tools (Claude, ChatGPT, Cursor) rather than feeling like I need to build agents. Areas in which I want to code against #AI APIs: - task repetition for multiple files - critique loops to improve text - pasting between non-MCP services

Started this great article about AI coding, and I already feel seen. Templates were so exciting in 2002. fly.io/blog/youre-...

I quite like paying for all of my AI subscriptions monthly. Every time I see the payment go out, it's a prod to remind me whether or not I'm actually using the thing. And if I'm not, then I cancel it.

Hilarious ChatGPT fail I came across: they cached the request too, so retries also failed 🤦‍♂️ Hype jockeys take note: this tech is still experimental, and can be completely unpredictable.

Decided I need a different outlet than LinkedIn for social chat. I was posting everything to everything, but I think that doesn't really work. I think I'm just going to use Twitter and Blue Sky for chat, expect much more ad-hoc posts here, with perhaps more nonsense (you've been warned)

Wrapped up the webinar on Prompt Stacking. That was fun! Lots of Q&A and chat. Sign up to my newsletter to find out when the next one happens: chrismdp.com

Planned out today's 2pm webinar: I'll be showing how to use Prompt Stacking to manage ever more complex codebases and prevent agents going off the rails. We'll look at: - The layers of direction (input, RL, hidden, agent, rules, instructions)

DX reports a 1.2x AI productivity boost, but teams are not shipping faster 🤔 I have been mulling their report and honestly it is a bit disappointing. It also felt incongruent to my own experience - but then I thought more abut what's really going on with AI coding.

I'll be at London Tech Week next Wed 11th! Looking for great conversations on AI-accelerated coding/productivity and building agents. Message me direct and we can grab coffee.

Unable to attend the Friday webinar? Register and I'll send you the recording. chrismdp.com/webinar

Get past the awkward start of AI delegation and unlock real leverage: just start doing it. You will not find a perfect plan. You just need to begin. More: chrismdp.com/unlocking-r...

“The next action should be the next physical, visible activity that will move the project toward completion.” David Allen (GTD) didn't know how important this would be in the AI age.

𝗬𝗼𝘂𝗿 𝗔𝗜 𝗮𝗴𝗲𝗻𝘁 𝗷𝘂𝘀𝘁 “𝗶𝗺𝗽𝗿𝗼𝘃𝗲𝗱” 𝗵𝗮𝗹𝗳 𝘆𝗼𝘂𝗿 𝗰𝗼𝗱𝗲. It looked good. Tests passed. You shipped. But two days later… bug reports start rolling in.

How can we make AI agents thrive in complex codebases? Join my next live coding webinar (Friday 6th June 2pm) to see the secret in action: chrismdp.com/webinar

What’s the collective noun for AI agents? Fleet? Workforce? Army?? (Not feeling great about that last one.)

Warning: If we never unplug, we’ll forget what we’re building for Last weekend, I got FOMO because I wasn’t directing more than one AI agent at once. That’s when I realised it was time to step away. Luckily I'd already planned to do just that :)

How to unlock deeper answers from AI: the two model trick If you want richer, more nuanced answers from AI, you need to go beyond the default. Most technical leaders ask a question, get a single answer, and move on. That is a missed opportunity.

Thanks @chrismdp.com for including @trustgraph.bsky.social in his recent discussion on GraphRAG! 🙏 www.chrismdp.com/graph-rag/

The future of AI is being built right here in London. 🚀🤖 ​Thanks everyone for coming to LLM London last night! 🙌​ (forgot to take photos - so here's an AI reconstruction) 😁

“Context is not a luxury. It is the foundation of next-level AI.” Find out more about Graph RAG after my latest deep dive, when I realised how much we need to improve and understand context. How we store and connect knowledge is the real unlock for generative AI.

A huge list of all consumer AI tasks and the tools and pricing for each option: www.chrismdp.com/ai-tools-list

How to build a solo business without trying to “go big”: Bees (yes, bees) help you to stop chasing mass markets. The more I study niche markets, the more I see the same truth: Mass appeal is overrated. Especially for bootstrapped entrepreneurs.

Thanks everyone for coming along to the webinar today! We successfully created a full multi-step complex change in Cursor, staying in control all the way.

Looking forward to live coding on this webinar at 3pm today! Unexpectedly free? Come along! We're doing two levels of coding: the strategy & planning ("making the change easy") and the execution ("making the easy change"), both with AI.

Smart from @cursor_ai. the IDE's days are numbered, and Cursor now has a background agent. Independent coding agents are the next big leap in code.

"#1 rule for a great tech event? Fantastic speakers with brilliant talks." Check ✅ Here's what we've got lined up on the 27th: LLM London Sessions #6 is happening: and it is going to be a good one. We’ve got 3 standout speakers, each doing something wildly different with language models:

I predicted AI coding agents would change everything. I said it would take a year. It took six weeks. I just tried Jules, a new coding agent from Google, one of a slew of mind blowing announcements at IO this week.

Google's AI servers feel massively under load today. Everyone has picked themselves up off the floor after that incredible set of IO announcements, and is trying everything out all at once.

Tomorrow afternoon, we're going deep. Starting from 3pm, we'll be using change documents to safely navigate a larger change on our AI Cheatsheet Generator app.

I found a brilliant tool that uses AI to power surveys for newsletter signups. No more guesswork, hello smart segmentation 💪 More below: Building an audience is challenging enough without worrying about sending the right content to the right people.

Can't make Friday's webinar? Sign up and I'll send you the recording: chrismdp.com/webinar I get it, things are busy. There's so much content out there. You can't be at everything. But you STILL want to know exactly how to manage larger changes within an existing application...

In his latest newsletter, @thebenedictevans.bsky.social hits the nail on the head: all prompt engineering should be abstracted away. This is exactly why I built kaijo.ai: to take away the pain of prompts and ship reliable agents that work. First for myself, soon for others.

“What’s that you say? A webinar that shows me how to safely vibe code larger changes in an existing application? On Friday 23rd at 3pm? And I’m free at that time?? Well sign me up!” chrismdp.com/webinar

If you want to get better at AI-first coding, learn to manage your context. The stupid bug that taught me this lesson:

Curious how to code a new feature AI-first for an existing app? Come to my free webinar next Friday, May 23rd at 3pm 💪 chrismdp.com/webinar

Earlier this year, on one fateful day, I deployed a chunk of AI-first code that looked perfect. It was not. When I first started using AI for coding, I fell into the vibe coding trap: accepting code that looked right without proper scrutiny.