Helping thousands of creators earn a living with HeightsPlatform.com. Founder of @LatchLoopAI @HeightsPlatform & @Velora. Host of The Creator's Adventure.
Austin, TX
- Tweets 3,507
- Following 408
- Followers 1,620
- Likes 8,310
Pinned Tweet
Introducing realtime voice in @LatchLoopAI
Navigate and control coding and knowledge work agents with just your voice.
Here's a video demo recorded with screen recording software that LatchLoop built for me.
Realtime voice control works on the mobile version too!
3
13
2,443
If you are excited to try OpenAI Dots right now, including the multiplayer team collaboration in a platform that works with any model and provider, LatchLoop is available to everyone now.
My team has been working this way since Feb 2025.
Free to start.
latchloop.com
2
1
8
264
So now I’m on pace for 1 million lines in the last 30 days actually.
Absolutely crazy times.
I'm on pace to deploy more than 1 million new lines of code to production this year.
And that is if models/harnesses don't get better, which they will.
What GPT 5.5 can do is a definite step change compared to the end of last year.
Absolutely crazy to think about.
1
113
I try to remain relatively relaxed as a founder, but I noticed if my agents haven’t done a couple hundred million tokens of work overnight, then I feel anxious.
I know, all I need is one more agent to be my proactive token anxiety agent and ensure there is enough work scheduled.
3
2
246
This would have been more than a year's worth of work for me pre-AI.
Now this is what happens in one repository in one month.
But this only represents about half of what I got done in the last 30 days across projects.
How its going:
LatchLoop made over 4800 commits and 443,000+ lines of code changed to itself in the last 30 days.
1
1
135
Merged a 363,000+ line PR this morning. Thanks Astra.
Wild times to get what would have been a year's worth of work done half way through a week.
2
67
We’ll look back on Astra’s release as the day AGI felt really here.
But we’ll also look back on the release of gpt-4 that showed the sparks that these models would reach this point.
GPT-6 Astra, trained on ~100K+ NVIDIA Grace Blackwell NVLink72. From ChatGPT to o1 to Astra in 4 years.
AGI has arrived. Congratulations @OpenAI team.
400K GPUs coming online next.
1
2
176
Tips for getting the most out of Astra:
1. It still doesn’t know how good it is. Don’t trust the first pass is the best it can do.
2. Avoid subjective requests. Don’t say, “Make this the best”.
3. Instead provide verifiable goals of what “best” looks like.
4. If you don’t know what could make your request verifiable, ask the model first!
8
208
This is insane what Astra can do. I had GPT-6 build this while I ate breakfast this morning.
Such an impressive result for almost no work on my end.
Astra is amazing. It built this game in 2 prompts.
One for the base game and one that mostly said "make the graphics better".
This is built with three.js running in the browser at 120fps.
1
4
467
5.6-sol is giving me all kinds of errors across runs. Who else is seeing the same?
Compute capacity is being moved over to prepare for Astra.
Astra is likely releasing Thursday based on this and recent press.
2
248
This is how my team and I have been working since Feb 2025 with multiplayer cloud agents for coding and knowledge work @LatchLoopAI
Two months ago, we started the mission to “build OpenClaw with OpenClaw,” and bit by bit, we moved everyone from using their local coding harness to using team.openclaw.ai - our shared agent that knows what everyone’s working on and orchestrates it all.
Multiplayer coding + infinite compute with nodes and cloud sessions has been a game changer for how we build.
Local harnesses feel like relics of the past now.
3
162
Model tier list based on practical usability and price/performance.
For the glm-5.3 I mean the flash model.
Fable near bottom of C because its simply too expensive for anything but occasional planning use.
Opus 5 at bottom of D because outside of 3D games it is not fun.
1
134
