Merlin

We gave our agents the ability to complain to us. At Rex, we’re running agents to automate order to cash processes end to end. One way to improve them is to read through their reasoning traces (the inner monologue of the agent) and see exactly why it chose one action over another. So we tried an experiment. What if we allowed our agent to send a complaint to us while working on a task? It started telling us about context it was missing, actions it wanted to take but couldn’t and raised questions about how to handle a situation. Whenever this happens, we get a notification in Slack. Most of the time, it points towards a direct improvement we can ship to make our agent better. A good principle seems to be to lean into the natural inclinations of LLMs. In short: Treat your agent like an intern.

@founderleague @eightsleep help me get to 100?

Certified professional sleeper. Congrats to our Week 1 Sleep Champion: @merlinkafka!

@sriramk @conductor_build checks a bunch of these boxes.

would love a new desktop agent super app - lets me switch harnesses and models and multiplex across them - makes it easy to move memory and context - can orchestrate between models ( use Fable as a planner but a lower cost model for daily driver ) - can retroactively look at usage and optimize for cost / better results

@tkkong Interested

Phil and I are hosting a watch party for the USA world cup match this Wed at our HQ in SF Marina. Engineers from OpenAI, Anthropic, and other frontier startups will be there. Few more spots remaining. Reply/DM for invite!

Just moved to San Francisco for @ycombinator (S26) 🌁

Most startups wait until the first big deal stalls to add SSO and SCIM. @rexdotinc did the opposite. Founded in January. Enterprise-ready in weeks with @WorkOS. And @Vercel AI Accelerator winner by April. This is how fast AI startups are moving now. workos.com/blog/how-rex-w…

@benln London

Cafe Cursor, round two. Where should we go first? • Paris • London • New York • Bangalore • Madrid • San Francisco • Berlin • São Paulo • Bogotá

So proud of our 2026 @vercel AI Accelerator cohort! 🚀 Congrats to @merlinkafka, @zeyu1337, @laurenlself and teams on winning!! And of course thank you to all our amazing partners: @awscloud, @OpenAI, @AnthropicAI, @browserbase, @NotionHQ, and more. Check out the recap ↓

We won Vercel's AI Accelerator - and are coming out of stealth @rexdotinc 🚀

@rexdotinc is out of stealth - and we just won first place in the @vercel AI Accelerator (out of 2500 teams applying) 🔥

@bnj Would love to stop by for an afternoon!

Slight preference towards other startups Particularly those that need design help

We’re a team of 5 with over 4500 sq ft of (beautiful) office Right next to Notion, Figma, Anthropic, etc If you’re a dreamer pushing the frontier, we’d love to host you in our space for free

@Jason I'm in

We started an AI founder twitter group... reply with "I'm in" if you're a founder and want to be added

Prediction: The next 12-24 months, "UX-pilled" builders will be in massive demand. Who can create intuitive interfaces, web+mobile+desktop apps that "feel good," natural, fast, and far better than the competition. THIS will be the difference vs those building "just" with AI.

After 6 weeks of building, Vercel AI Accelerator teams are ready to show their ships. Join us in SF on April 16 for Demo Day, where founders will pitch to @Vercel, @AWSstartups, @AnthropicAI, @OpenAI, and more. Save your seat ↓ vercel.com/go/ai-accelera…

@rahulgs How do agents learn tribal knowledge?

seems obvious but: things that are changing rapidly: 1. context windows 2. intelligence / ability to reason within context 3. performance on any given benchmark 4. cost per token things that are not changing much: 1. humans 2. human behavior, preferences, affinities 3. tools, integrations, infrastructure 4. single core cpu performance therefore, ngmi: 1. "i found this method to cut 15% context" 2. "our method improves retrieval performance 10% by using hybrid search" 3. "our finetuned model is cheaper than opus at this benchmark" 4. "our harness does this better because we invented this multi agent system" 5. "we're building a memory system" 6. "context graphs" 7. "we trained an in house specialized rl model to improve task performance in X benchmark at Y% cost reduction" wagmi: 1. product/ui 3. customer acquisition 4. integrations 5. fast linting, ci, skills, feedback for agents 6. background agent infra to parallelize more work 7. speed up your agent verification loops 8. training your users, connecting to their systems and working with their data, meeting them where they are

@cjpedregal @meetgranola Massive! Congrats

Today we're announcing our Series C alongside some big updates that make @meetgranola better for your team and your tools. Excited to partner with Danny at Index and Mamoon at KP. Big things to come. Back to work!

Incredible cohort. Excited to be part of this and share more about what we're working on soon

@karimatiyeh Exciting. Signed up

I grew up in Lebanon. My wife is Swiss. Half my family is French🥖 + Swiss🫕 and I spend a lot of time in Europe. Europe has incredible builders — many of them work at Ramp. But building companies there is still far more painful than it should be. I’ve always wanted to see Europe move as fast as the US. Yesterday we announced Ramp’s expansion into the UK and EU with the acquisition of Billhop. Let’s help make that happen.