Pinned Tweet
British Airways has the worst app and your loyalty anyway. The moat was never the software. It is routes, slots, and the fact that you need to be somewhere. AI improves the part that was never why you stayed.
1
174
Nida Sahar retweeted
We take our mission to help build a better Internet seriously. The Internet needs us now more than ever. We’ve got its back.
Dudes @Cloudflare is doing more to secure the internet than any other company right now.
And most of it is free.
most small teams run their whole crm in a google sheet anyway. wrote about what changes when your ai agent can finally use it too: dev.to/nidasahar/your-crm-is… #MCP #AI
1
56
Nida Sahar retweeted
I get so excited about this next generation of compute. Just so cool.
boxd in 60 seconds
a whole linux machine that boots in under 10ms
composable: set it up once with your tools and data, then spin up as many as you need
fork it mid run, let it hibernate, wake it on traffic, share it with your team
one line to install: curl -fsSL boxd.sh/downloads/install.sh | sh
Nida Sahar retweeted
I included a few references to this year's record-breaking Kākāpō breeding season in a talk I gave yesterday, and since Claude Opus 5.5 is surprisingly capable at pixel art animation I had it create this celebratory video for my closing slide
Nida Sahar retweeted
One config file took down the internet last November, and the code that crashed was working exactly as written.
That was Cloudflare's worst outage since 2019. A routine database permissions change made a query return duplicate rows, a config file doubled in size, and the proxy hit a hard limit nobody was thinking about.
The code panicked and started throwing 5xx errors, and because the file regenerated every five minutes, the network kept flipping between fine and broken. It took about three hours to trace it back to a config file.
The code did exactly what it was written to do. It just met data nobody planned for, and that only shows up when the thing is running.
Here's why this matters more now that we're all shipping faster with copilots. Code gets written quicker than it gets read, and it's judged on whether it builds, not on how it behaves with real traffic. Add an LLM and you get new runtime unknowns too.
Florian Mair has a good post on the @Dynatrace blog about exactly this, why runtime is where code tells the truth: fandf.co/46e5Wx7
Their tooling is built around closing that gap. The MCP server pulls live runtime context (errors, traces, slow queries) into your IDE, and the AI Observability app captures prompts, model calls, token usage and downstream behaviour in a single trace. There's a free 15-day trial, no credit card, if you want to try it on your own stuff. Sponsored by Dynatrace.
🤝 Paid partnership
Nida Sahar retweeted
"GKE ambient networking provides a simplified, sidecarless deployment model for a service mesh with Layer 4 capabilities."
Neat. The service mesh was always an advanced component in a platform. It needed an easier approach.
@GoogleCloudTech Docs: docs.cloud.google.com/kubern…
Ten AI agents in a simulated town invented slang nobody taught them. 'Ledger remembers who' meant your actions stay on record. Sixteen days in, they had office jargon. The researchers read it all and missed half. Intelligence gets strange the way teams do: an inside joke.
1
52
Blue-green deployment is two full production systems and one honest admission: the new version has not earned trust yet. The idle half looks like waste every day except the one day it is everything.
1
29
The models are ahead of the habits. The catch-up will not look like adoption headlines. It will look like one boring workflow rebuilt at a time.
Even if AI development stopped today, we'd have years of catching up to do. The gap between what current models can do and what almost anyone is using them for is vast.
Here’s my post on The Overhang, and the four advantages that let people close it. open.substack.com/pub/oneuse…
23
Refusing to find it interesting stopped being a personality trait. The ground moved under the field; curiosity about where it went is now part of the job.
Being a computer scientist who refuses to find anything about LLMs interesting right now is a bit like being a geneticist who refuses to find anything interesting about the recently opened Jurassic Park
34
Try, fail, then make a film about the attempt. An uncomfortably accurate summary of how research actually goes.
Hey Claude, "Pick a problem or mystery that obsesses you and solve it as best you can & make a movie we can share on social media about it"
So it took a crack at the Voynich Manuscript & failed. Then it made this movie, which is pretty interesting to watch and a good explainer.
18
A repo of instructions for agents just passed React in stars. The most influential code of this era might end up being prose.
Pretty incredible how the skills repo by @mattpocockuk is only 6 months old, and has more stars than the React one. We had to sit down and talk. Timestamps:
00:00 Intro
05:48 How Matt got into tech
10:14 How Matt got into open source
12:58 Joining Vercel
18:39 Total TypeScript
23:21 AI’s impact on technical education
30:32 Building reusable skills for AI coding agents
40:46 The “smart zone” vs the “dumb zone”
45:02 The wayfinder skill
47:52 Why agents excel at software engineering
50:54 “Leading words”
1:01:10 Learning the fundamentals
1:09:17 Local vs. cloud agents
1:12:36 Planning vs. course-correcting
1:18:13 TDD and agents
1:23:06 Living in the UK
1:24:21 Teaching: the human part
1:28:36 Advice for junior engineers
1:31:07 Gardeners and great engineers
1:34:01 Book recommendation
Brought to you by:
• @turbopuffer – vector and full-text search engine built on object storage. It’s fast, cheap, and extremely scalable. turbopuffer.com/pragmatic
• @linear – the product development system trusted by 40,000+ teams to plan and ship. Get six months of Linear Business for free: linear.app/pragmatic
• @WorkOS – The fastest AI-native teams have to slow down for the hard problems. WorkOS makes sure auth, for your app and your agents, is never one of them. workos.com/
Thanks for the convo @mattpocockuk!
17
Funny timing for this question. Design docs matter more now, not less. The agent implements what the team wrote down. Everything else it guesses.
What does a *really* good software design doc look like?
We can argue about when you need to write one, how much to put in there, and even who reads it nowadays, but it's hard to argue that these are the areas that deserve coverage.
refactoringenglish.com/excer…
2
41
1/ Someone new joined a team channel last week. I watched them type a question, delete it, type it again. A short thread on what seniority is actually for.
1
28
4/ The strongest seniors I know aren't the ones with the best answers. Their channels are just full of unfinished questions.
1
9