Big data and streaming analytics and cool software building nerd. Retweets != endorsements and opinions given are my own, even though you should agree with them
Joined March 2013
- Tweets5.5K
- Following219
- Followers133
- Likes1.2K
Pinned Tweet
I'm trying to find a sister I was recently made aware of.
tiktok.com/t/ZT6anwuv2/
👇🏿👇🏿👇🏿👇🏿
Here is the trap I am watching teams walk into with AI coding agents.
Someone points an agent at a service, and in an afternoon it generates more code than the team would have written in a week.
Then the next three weeks are spent quietly discovering that the agent dissolved the boundaries they had spent years defending.
It did not do anything wrong but It did the only thing it could because It had no idea why those boundaries existed.
That reason was never written down anywhere, it lived in the heads of the people who built the system.
This is the uncomfortable heart of harness engineering, all the scaffolding we build to steer these agents is really just one thing: encoding the context the system needs so it produces work you would have signed off yourself.
And that turns out to be genuinely hard, because articulating the why behind a distributed system, clearly enough that something without your context can act on it, is a skill most of us never had to practise.
We always had humans to fill the gaps.
Garbage context in, garbage architecture out but now at machine speed!
Getting good at this is the whole reason I facilitate the architecture cohort for InfoQ, where senior engineers are made to explain their real decisions to peers who do not share their assumptions. It is the exact muscle AI now demands.
Next cohort starts 13 August, link in the comments.
#architecture #distributedsystems #ai #softwarearchitecture @InfoQ
Ed Brown retweeted
Anthropic engineer:
"You're not supposed to prompt Claude. You're supposed to build a system that prompts itself."
In 45 minutes, she breaks down how Anthropic builds agents that remember, learn from their mistakes, and get smarter with every run.
Worth more than any paid course you'll find on building agents.
Watch this and bookmark
Ed Brown retweeted
🚨 Anthropic just showed a 27-minute workshop on how to actually do prompts for Claude.
Taught by the people who built it.
Free. No registration. No paywall.
I've seen $300 courses that don't cover what they teach in the first 8 minutes.
Watch the session
Ed Brown retweeted
People now try to steamroll my parents since they're mildly elderly. So part of my daughter duties have become popping in to get folks back in line when they're pushing it a little too much lol
The role reversals in parent/child relationships continue to crack me up.
Ed Brown retweeted
I tried to use X Ads, really did
(because via the subscription I get $1,000 of ads credits p/m)
It's a product I failed to use properly, because it's just so bad, limited etc
So a tweet celebrating having a skeleton crew run a product I cannot use is not the flex it looks like
Ed Brown retweeted
Bleacher Report gave the Ravens a "D" grade for their 2026 offseason report card
"It's difficult to look at the Baltimore Ravens and say they're a better team today than they were last season"
Ed Brown retweeted
This is why I got rid of my Oura ring years ago.
I’d wake up feeling absolutely amazing and well rested. Only to check my phone and have it tell me I actually slept like crap and should take it easy today.
I then found myself not working out in case it was right.
This quoted post is unavailable.
Ed Brown retweeted
Amazon is holding a mandatory meeting about AI breaking its systems. The official framing is "part of normal business." The briefing note describes a trend of incidents with "high blast radius" caused by "Gen-AI assisted changes" for which "best practices and safeguards are not yet fully established." Translation to human language: we gave AI to engineers and things keep breaking?
The response for now? Junior and mid-level engineers can no longer push AI-assisted code without a senior signing off. AWS spent 13 hours recovering after its own AI coding tool, asked to make some changes, decided instead to delete and recreate the environment (the software equivalent of fixing a leaky tap by knocking down the wall). Amazon called that an "extremely limited event" (the affected tool served customers in mainland China).
The new "rm -Rf *" from root
Claude Code wiped our production database with a Terraform command.
It took down the DataTalksClub course platform and 2.5 years of submissions: homework, projects, and leaderboards.
Automated snapshots were gone too.
In the newsletter, I wrote the full timeline + what I changed so this doesn't happen again.
If you use Terraform (or let agents touch infra), this is a good story for you to read.
alexeyondata.substack.com/p/…
ALT A console screen shows a critical warning about a destroyed database and missing backups, outlining deleted infrastructure components.