building things at @SpiceLabsHQ · managing @KlawiterDesign · mass-deploying agents into the void 🏳️🌈
Los Angeles, CA
Joined May 2008
- Tweets3.9K
- Following223
- Followers198
- Likes113
Ryan Brodkin retweeted
Things you can use Al for when working on UI:
- ask it to try and break the stuff you built
- let it create state machines to check different scenarios (success, error, different user roles, etc.)
- when finished, ask it to make before/after screenshots
- make it ask you lots of questions about the thing you are working on to validate the idea
- let it create a few variants of a Ul piece so you can try it out
- setup loops to maintain a quality bar by constantly monitoring
Used 5.1 from almost the moment it dropped today. We may be reaching the point where performance gains may only be noticed in specific domains or benchmarks at this point, but the lower costs are very welcome. Good job, @AnthropicAI!
30
Some people forget that a year ago we had basically nothing. The fact we have usable AI at all in this short time is a testament to AI itself. While I agree there are improvements to make, the pace of advancement is incredible. Meanwhile, Microsoft is sitting on 15yo bugs.
Anthropic will say things like, “Loops write 30% of our code,” and then burn an unfathomable number of tokens just to ship remarkably mid software.
The models are amazing. Claude Desktop and Claude Mobile? Genuinely bad. And apparently increasingly loop-generated.
It all feels backwards. It feels like a strategy to increase usage because loops can run 24/7. Anthropic benefits from the increased token usage, while the rest of us get more slop.
And I don’t just mean code slop. I mean product slop, design slop, prose slop. Another calorie-tracking app. Another landing page with *that* aesthetic. Another article that uses the term “load-bearing.” The progressive convergence of the web toward the mean.
No one wants this. We want less software that’s more thoughtfully crafted. And I don’t think we’re yet at the point where looping agents are going to give us that.
Or maybe I’m just a Luddite.
1
40
Ryan Brodkin retweeted
Strongly agree with all of the above. We are entering the next era of code, where the model is able to generate correct code for an increasingly large percent of tasks.
Our job is to make sure the model and our systems have the right guardrails, then to run Claude Code + an advanced model + a verifier in a loop and feed it tasks (or, give the model the data it needs to generate tasks), finding and getting rid of bottlenecks along the way.
The UK Online Safety Act requires "highly effective" age verification but leaves the method totally open. Predictably: Pornhub is demanding government IDs, Spotify is running face scans through a third-party vendor. Every platform building its own identity pipeline.
The UK is on a dark path for speech on the internet.
I argued in the Telegraph in December that laws claiming to be for child safety were opening the door to surveillance and censorship of law-abiding adults. The new ban continues this trend.
1
194
California went a different direction. AB 1043 requires the OS to collect age at device setup—iOS, Android, etc. and expose a bracket (under 13, 13-15, 16-17, 18+) via API. The app gets a signal. It never touches your ID. One chokepoint, privacy by design.
1
1
38
The UK law isn't wrong about the goal. It's wrong about who should hold the data. When you leave the implementation up to every platform independently, you get a thousand different ID vaults and surveillance vectors. California figured out the architecture [QT: @cb.substack.com]
8
Has he not seen his approval ratings? If people are mostly anti-Trump and we assume, for a moment, that judges are people, then most judges are anti-trump by extension
President Trump fumed over the order to remove his name from the Kennedy Center, calling the judge responsible "an anti Trump Hater" in a lengthy post on his Truth Social platform. abc7.la/SorAeE
1
14
This, kids, is why California requires servers to be paid a minimum wage in addition to tips. Some restaurant owners think it's OUR job to pay their employees 🙄
5
Ryan Brodkin retweeted
Asking two agents in different harnesses to debug your code
(from andirockk on IG)
Ryan Brodkin retweeted
Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D. I remain deeply passionate about education and plan to resume my work on it in time.
Ryan Brodkin retweeted
ANTHROPIC beats OpenAI in business adoption for the first time. per @tryramp data
Today's update of Ramp AI Index shows 34.4% of businesses using Anthropic versus 32.3% using OpenAI.
Adoption of Anthropic quadrupled over the last year, while OpenAI rose only 0.3%.
Ryan Brodkin retweeted
Effective today, we are:
1) Doubling Claude Code’s 5-hour rate limits for Pro, Max, and Team plans;
2) Removing the peak hours limit reduction on Claude Code for Pro and Max plans; and
3) Substantially raising our API rate limits for Opus models.
Ryan Brodkin retweeted
do you understand what just happened to Robinhood..
Someone sent a perfect phishing email - real domain, DKIM pass, SPF pass, DMARC pass and Robinhood's own servers delivered it.
Here's the chain:
→ Gmail treats john.doe@ and johndoe@ as the same inbox
→ Attacker registers a NEW Robinhood account using the dot trick of YOUR email
→ Sets the device name to raw HTML code
→ Robinhood's "unrecognized activity" email renders it unsanitized
The "Review Activity Now" button? Attacker's phishing site.
The email? 100% real.. Sent by Robinhood.. Signed by Robinhood..
Just because it passed every security check doesn't mean it's safe.
New Robinhood phishing chain that's kinda beautiful:
1. Attacker creates an RH account using the Gmail dot trick of your email (same inbox, different address)
2. Sets device name to HTML
3. RH's "unrecognized activity" email renders the device name unsanitized (html injection)
The result is a real email from [email protected], DKIM pass, SPF pass, DMARC pass, with a phishing CTA
Just because it's real, doesn't mean it's safe... $HOOD
It's honestly not nasty enough! Add a frameset, make lots of gaudy buttons with poorly rasterized text, list which browsers your site works on (Only IE and Netscape, obvi)
10