We build the AI systems companies actually run on, and teach their people to run them. Strategy, design, and technology, side by side. Let's cook.
Louisville/Cincy/LA/NYC
Joined February 2026
- Tweets58
- Following88
- Followers52
- Likes108
Pinned Tweet
This is the #1 reason we built Roux. And just yesterday we found ourselves again saying, "we should send that team a Roux to get a sense of all that the product needs to solve for". buff.ly/R5f8veW
Conference evals get a lot of "it was fine."
Roux asks the follow-up: what would have made it worth the trip? Every answer updates one live report, so your team sees patterns while the event is still running.
Built by Gumbo. Try it free: helloroux.com
Q4 is when every team gets asked "what do you need next year?" in six different meetings.
Send one Roux. It asks each team the follow-ups and you get a single planning brief back before the first budget review.
Built by Gumbo. Free to try: buff.ly/Vgo3oIt
Forward-deployed engineers can do more than ship a feature. Working inside Dermsquared’s team, we helped move its conference app and learning experience forward together. We built a year-round customer experience that doesn't just stop when the conference ends.
Big fan of this partnership and looking forward to the next era of building for Dermsquared: hellogumbo.com/case-studies/…
Project handoffs usually take a few meetings to get right.
Send a Roux to the person rolling off. It interviews them on what’s done, what’s half done, and what never got written down.
The next owner starts with a full report, so the one meeting you do hold can be a short one.
Inspired by @mattpocockuk’s /handoff: buff.ly/mEuv4Mx
Built by Gumbo. Try it out for free: buff.ly/QwO86mG
Budget owners:
2027 planning is about to be a lot of "just circling back." Do you really have time for that?
Roux interviews every team for you and turns it into one budget brief. Fewer meetings, same answers.
Try it for free and let us know what you think!
buff.ly/Vgo3oIt
One of our fave dynamic duos
Hackathons with @stevecaldwell are my favorite excuse to try something new. The last one turned into Roux for ChatGPT and Codex, and I'm kind of stoked on how the page came out. Check it: helloroux.com/chatgpt
Your AI has the data. Your team has the context. Roux connects them.
Add Roux to ChatGPT, send an AI interview, and get back what people agree on and what to do next.
helloroux.com/chatgpt
GUMBO retweeted
one sales call.
one transcript.
one chat with my @OpenAI Dot.
one Roux created via MCP.
one email drafted to prospect with the Roux link.
one hundred responses collected in 24 hours.
one proposal created.
one deal closed.
one happy founder (me).
helloroux.com/chatgpt
We got tired of endless meetings just to generate Granola transcripts for Claude, so we built a tool that captures human context async and synthesizes it for you.
Now you can use it too!
Roux interviews your whole team from one link and turns every answer into one clear picture.
Skip the meeting. Send a Roux.
buff.ly/WrCxFlu
Our team is always building.
i've spent so many tokens over the past few months building games in three.js, porting things to @UnrealEngine has made me realize how much of the wheel i've been reinventing. working in a proper game engine is gamechanging.
For those wondering what the heck Jev is, and how it works, we made a simple doomer demo to really demonstrate the ins and outs of Jev.
shouldaikillusall.com/
The Gumbo team builds. A lot. If you want to ensure you always see our latest experiments, make sure to sign up for our email newsletter: hellogumbo.com/newsletter
Last week's issue included our latest game, built in a weekend as a test of Astra's capabilities. You can test out Trailer King here, for free: trailerkingonline.com/
GUMBO retweeted
unsure where to start with @typesafeai Jev? bookmark this and point your agent at it for inspo.
github.com/hellogumbo/awesom…
GUMBO retweeted
Here's a 45-second TL;DR on Jev.
I find the core idea beautifully simple, but the video made it really hard to understand.
Hope you find it helpful.
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI?
I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev
• 20-200x faster
• 40-400x cheaper (w/ output tokens free)
• Frontier composable intelligence optimized for decisions
AFAICT the shortest path to AI-based economic revolution