🌊 Dev For Lola 👉Click the link below👇 👇
Joined December 2009
- Tweets14.4K
- Following8.8K
- Followers9.5K
- Likes907
Teaching Lola to keep englishTranslation: backstage, where it belongs. đźŽđź‡«đź‡·
Next act: smoother onboarding, natural voice conversations, and zero accidental JSON monologues. 🎙️
flystewie retweeted
after Opus 5.5 it actually hurts to look at anything GPT 6 Astra designs
gave both models the exact same prompt: "make a dynamic 15-second motion graphics video that shows what an incredible motion designer you are, like it's your showreel for a résumé. go all out."
not even close, see for yourselves. GPT is honestly terrible here, openai just has awful taste
flystewie retweeted
Opus 5.5, what are you doing?
It made another perfect jelly whale for $9.26. Translucent, squishy, and full of tiny bubbles. You can spin it, squish it, and toss it around while its whole body flops and wobbles
I just wanted to check the result. Now I can’t stop tossing this little guy around
The most expensive UI bug is making people ask AI the obvious next question.
Click a mayor’s bike policy? Show the council evidence, trade-offs and unknowns right there.
Save the chatbot for the question the interface couldn’t predict.
Toronto votes today. Still deciding? 🗳️
Five candidates. Six issues. One place to compare.
I built a free guide to explore the plans, trade-offs and sources before you vote.
Independent. No candidate endorsement.
toronto.lolalingochat.com
Toronto votes today 🗳️ (10am–8pm).
I built a free, nonpartisan guide comparing 5 mayoral platforms, with sources and a link to Toronto’s official voting lookup.
Independent project, not affiliated with the City.
toronto.lolalingochat.com
Most prototypes end with a “Sign up” button.
I think they should end with “Keep going.”
A game, travel planner, or election explainer should be able to hand an AI the unfinished question.
The demo can end. The conversation shouldn’t.
Adding a chat box to analytics is easy.
Getting the chart, the agent and the Slack thread to agree on which “this week” they mean is the product.
Context is a feature. Amnesia is expensive.
I want to click a number and ask my agent why.
But the answer needs the same date range, denominator and source as the chart.
Otherwise I’ve just given the dashboard a very confident mouth.
I hired five agents and accidentally became their middle manager.
PIP can test onboarding on a cheap Luna route; I just want the screenshots, timing receipt, and one decision in Slack.
If I have to read the whole run, I'm the bottleneck with a nicer title.
The fastest way to ruin a learning game is to reward the dog for understanding French.
Lingo gets the stars. The learner gets the evidence.
Different scoreboards. Very different product.
Built an AI team to ship faster. Somehow they started spending more time verifying the work than doing it.
Now Lorie has to audit the auditors đź’€
I learned building a language learning app
The model can sound like a great teacher while the learner still can’t order a coffee. I’m making the learner’s next conversation the test. A smooth demo is a very persuasive accomplice.
The real production test for an AI dashboard isn’t whether it loads.
It’s whether you can delete it and the evidence survives.
Otherwise you’ve just shipped a screenshot with API keys.
