🛠️ @stedi Prev: @Trek10Inc, @goserverless ☁️ AWS DevTools Hero. 🧔 he/him.

Washington, DC
This explains a lot.
BREAKING: the UK government publishes official rules on how to use AI: 1. Ask first if you really need to use AI 2. Check if a spreadsheet can do the job before using AI 3. Choose the worst model possible so it uses less energy 4. Keep prompts short to reduce the environmental impact 5. In general, use AI only when necessary, as a climate-saving measure With a mindset like this, we should accept the UK back into the European Union
244
This is an extraordinarily bad time to realize I actually need more RAM on a machine…
1
6
376
Removing plan mode doesn't mean you can't plan... the models are fully capable of going through that process with you without some top level flag. In fact, it's nice to allow them prototype / inject small bits of testing code during planning to rapidly confirm assumptions.
135
now it's just my idea for a solution was bad, and I can try again faster.
before ai you would - write some code - it works, ship it - few months later you'd realize it was bad - you'd refactor it - repeat not sure much has changed. you can produce a lot of code now but you can also refactor a lot of code
250
more in this space than not. i have SO many tiny utils now that I never would have done before because it wasn't worth the time. They work _exactly_ like I want at this moment in time, and if my needs change, it's just a prompt away.
everyone's arguing about the details of code mattering or not but you can have a lot of fun by working on things where the code definitely doesn't matter there's a mountain of things in this zone you never even considered before
1
255
the number of times I looked at code and thought "what absolute buffoon wrote this shit" and git blame was me... he right.
Best part of AI is shattering dev's egos, the worst part of this industry. You are not special. Your code actually sucked like everybody else's. You were never the smartest person. The things my sales guy BIL and designer brother are doing is beyond anything I have done.
2
1
519
Jared Short retweeted
Opus 5.5 is here! 🎉 It's honestly such a nice model to work with. Feels like Fable, but ~30% faster and ~40% cheaper per task than Opus 5. Go try it out!
302
232
5,035
290,495
at this point my purpose it to just toggle the models in the harness to the latest and greatest.
161
Also super effective for embedded firmware... I threw fable at an espressif chip and it decided to rapidly verify things by doing memory dumps to see what it was writing to a screen, rendering it to a PNG, and doing visual checks.
The thoroughness with which agents can test and QA software is unrivaled. I pointed out something wasn't rendering right in a mobile (in-app) browser. The thing goes to the depths of Mordor to reproduce, simulate, fix, deploy, verify. It created an ephemeral vercel deployment to throw at an iPhone simulator! The absolute grind on these things. Humans could simply never match this level of intensity. We'd run of energy, we'd extrapolate, we'd hope it'd work out. The software of the future will be of a quality and performance we've never experienced before. It's amazing to watch this shift.
1
411
i will not play the wow beta. i will wait for release. i will not play the wow beta. i will wait for release.
1
217
On the flip side… in many cases I only want 5% of the functionality of some service. I can customize and do that 5% the exact way I want without the overhead and cost of the other 95%. The trap is to only build the 5% I need and not keep building the world while I’m at it.
the hardest people to sell software to now are people who can build with ai they’ll spend a week building a shitty version of your app to avoid paying $20/month then maintain it forever
1
1
411
My biggest unlock for self-improving personal software is extensive tracing early. If it does something I don't like, a quick "look at my last trace, it did X instead of Y. Make this trace an eval, and iterate to a fix". Incredibly simple, but often the most effective things are.
2
1
7
529
Someone just referred to a video game system as being from "back in the 1900's" and I don't know what to do with my feelings about this.
3
265
Turns out "chat" maybe isn't the ideal interface to integrate AI into apps after all... wild.
203
Jared Short retweeted
There's been a ton of talk about the role of humans in code review, and when and how humans should be signing off on changes. I believe that, long-term, humans have no role in routinely reviewing code.
137
166
1,712
506,117
Help I'm running out of stuff to classify and still have $4 of credit left. Already labeled all my email and cleaned up my whole file system.
1
306
I think my takeaway from Jev so far is that latency matters a lot more than I was giving it credit for.
4
1
127
5,003
Turns out all I needed for Wispr Flow to fit into my workflow was a cheap Bluetooth mouse ring with a touchpad and my own interceptor software for gesture handling to quick‑switch apps and toggle the microphone.
1
429
The models are good enough now that I just one shot little utilities as native apps to fix all the random annoyances I have. Example, a "link router" that masquerades as my default browser and can point directly to native apps for any rule-matched URL, otherwise go to chrome.
1
184
Now I if someone sends me GitHub links OR linear links, I just get dumped straight into my preferred linear view of things. Don't even have to deal with a flash of chrome like linear links normally plague me with.
108