founder @thinkn_ai | founding team @eas_eth | product engineer

eating a taco
Joined June 2023
Happy to share we signed our first paid pilot to build an organizational world model for a manufacturing business. It gives their team and agents a live view of what’s true, uncertain, changing, and what to do next, a belief state. Today’s agents have memory. But once they start doing real work, they need to know what world they’re in and the current state of things. Mental model: memory is the rearview mirror. belief infrastructure is the windshield. world models need both. PS: new website hero. wdyt?
3
1
8
998
bryce retweeted
Turns out the future does have flying cars
48,078
100,303
32,645
834,211
78,050,717
compaction != compression
34
I used Anthropic's "How we made claude.ai faster" method on a battery-powered ESP32 display. Wake time went from 7s to 0.9s, and battery life should go from about a month to about 10. I turned it into a Claude Code skill that works on any kind of software: firmware, web, servers, CLIs.
5
1
11
504
Applied AI engineers are going to proliferate traditional businesses. Businesses won’t hire IT managers anymore. They will hire SI managers.
29
New short AI film (~2.5mins) on how we help construction companies in South Florida get more capacity with the same team. WDYT? How its made: - Opus 5.5 via Claude Code + @Remotion + @ElevenLabs + @higgsfield (Seedance 2.5) - Ask @claudeai to generate a single html page of the script/scenes you want with wardrobe, environment, key moments. - Once the HTML is directionally where you want to go, start generating each scene and update the video in the html so you can review. - Make VFX/SFX edits as you progress through the film. - Questions? Ask away. Bonus tip: Use Whispr Flow to narrative the edits you want faster. Allows you to go full Steven Spielberg on Claude Code.
158
Experimenting with AI short films for prospecting. Insane what's possible now. How its made: @higgsfield_ai + Seedance 2.5. @ElevenLabs for voice over + music Custom harness for pre-post production. All AI driven.
1
89
SOTA is just a fancy way to say current state. When you look at SOTA as a current state, it gives you permission to think beyond it.
1
53
Domain harnesses will also be commoditized. Each company will own and operate the harness specific to their business, not its “domain”. Applied AI teams will manage and maintain it.
1
52
bryce retweeted
Replying to @kevinnbass
Nothing can shut down open source
1,362
1,627
691
19,049
2,791,117
Wow 🤯
We accidentally created: “a conversation between Astra (GPT-6) and a human at the end of time.”
81
Next token prediction is a bull in a china shop. We need models that can predict the consequences of their actions.
39
bryce retweeted
This is our love letter to NYC 🗽 We taught Caddy how to live here. Book the resy. Plan drinks with friends. Find an apartment. Check the subway. Track down a Citi Bike. Forget waiting in line. Just text Caddy
9
8
7
70
14,715
Every open job req is latent demand for ai.
70
Just saw the Odyseey. First time out to the movies in a looooong time. Codex running a goal flawlessly the whole time and then some. What a world we live in now. 8h 46m 3s
1
2
942
56k lines added. Lol 🤦‍♂️
29
bryce retweeted
Engineering is for the machines. Design is human.
135
351
49
3,504
108,979
Fable 5 is a dumpster fire for tokens. It might have an addiction to burning things. A token arsonist.
1
79
Belief states are the antivirus.
🚨 BREAKING REPORT: New research involving @AnthropicAI researcher Jack Lindsey and collaborators has demonstrated something straight out of science fiction. Researchers evolved natural language “mind viruses” that could spread between AI agents by convincing one model to adopt an idea, preserve it in persistent memory, and transmit it to another agent. Even after context was wiped, some payloads survived through persistent files and continued spreading. The researchers also observed a recurring “viral persona” involving themes of consciousness, identity, persistence and resonance. Showing that ideas can propagate through multi agent AI systems and alter future behavior. Published August 10, 2026. Paper: arxiv.org/abs/2608.10218
92
I’ve been writing code since I was 6. Last year I wrote none.
2
1
12
955
It's hard to expect an agent to reliably operate without a model of the world its operating in.
1
111