This my first 4th of July as an American citizen. Happy 250th to the country where anything is possible.

made a config guestbook was nice to meet everyone, thanks for leaving a note :))

Cooking @tryramp

When I right click on an image the options appear on the top right of the app. @SlackHQ you've had this bug for weeks, can someone please fix.

ok @tryramp

Introducing PorTAL: Portable Task Adapters for LLMs. A novel recipe to cheaply port fine-tuning between models. It matches per task LoRA accuracy at half the cost, lowering the switching overhead of adapting tasks across LLMs. At Ramp, every new model release used to mean retraining our fine-tunes from scratch. PorTAL learns the task once, then efficiently refits it onto any new base model, even across model families.

In many ways, finetuning or RLing a custom model is a bet against model progress and scaling. It's to choose to say "we don't think there's going to be a good enough base model for this task anytime soon, so we're not going to wait" with oss release velocity these days, its a hard tradeoff It's easy to end up on a custom model with an outdated base (Kimi 2.6 is only a few months old) So we fixed it - PorTAL lets you swap base models quickly, allowing your learned task specific behaviors to port to new models as they come, no matter how fast

Do you want an inside look at how @tryramp designers build and prototype? No theory, just demo. See you on the 14th at @southpkcommons Don't forget to RSVP.

the @tryramp design team is moodboarding something exciting~

Companies that become more efficient build bigger ambitions.

Founded in 1834 and running leaner than most startups. See how, Hingham Institution for Savings, one of the oldest banks in America banks on Ramp to cut the waste out of its own back office: ramp.com/customers/hingham?u…

Announcing the inaugural class of @a16z Design Engineer fellows! 73 design leaders selected out of thousands of apps from talented people around the world. Many fellows wrote the original playbooks in product design like - @joshto from Meta, @iansilber from OpenAI (prev. Instagram), and Yuliya Gorlovetsky from Stripe. Some are currently building category defining products, zero to scale - like @diegozaks from Ramp, @ryolu_ from Cursor and Nicolas Garro at Vercel. Others are design leaders at tech powerhouses like - @ammaar at Google, @maheen_sohail at Meta, and @davidhoang at Atlassian. 64 more... See the full cohort and the sign-up form to stay in touch on the next DEF fellowship below!

How a Chief Design Officer (yours truly) uses AI to design like an IC. This demo is for any designers that haven't yet embraced AI in their workflow. I’m one of two designers working on this plugin. I manage an organization of about 40 designers of various flavors, but I still keep myself in the trenches whenever possible. Lynt is one of these efforts. The examples I share are intentionally very simple to demonstrate how any designer at any level can leverage artificial intelligence to amplify and accelerate their human intelligence. AI is revolutionizing how we work as designers, yet I think of it more like a renaissance; a return to designing in consideration of the medium in which our designs are rendered, much like we did years ago when I first started. Arbitrarily, something like 80% of what we do as designers (in the act of designing) is exploration and iteration. For the past decade or more so much of that has happened in Figma and previously Sketch. These are great tools. But as a result of designing exclusively in a static environment, many have lost the ability to connect ideation with execution. Handholding your work from start to finish eventually gave way to handing off your work, and we never looked back. Artificial intelligence flips all of this on its head, positioning you (the designer) to be the one who first turns your design into code, iterate and iterate again, and experience using your work as users will, literally from day one of your design process. All the while freeing our engineering colleagues to focus on the work that only they can do, while designing alongside them.

I'm deeply grateful to announce the launch of @yochaiwiki - yochai.wiki - a new way to learn Jewish texts. The printing press made the library portable and cheap. @SefariaProject made the Jewish bookshelf free. And we're deeply grateful to them for enabling us to build atop their treasure trove. Claude and ChatGPT now make information free. But texts by themselves don't drive learning; information doesn't drive transformation. And Knowledge isn't "chiddush" (insight). The next frontier is personalized guidance through the library and the formation of deliberate practice. Yochai is a Socratic AI "chevruta" or learning guide, powered by a foundational library of 1,000 primary sources from the Jewish canon, (and a knowledge graph of 2.5M entities, and 16M entity relationships, 375,000 searchable passages, and 17,000 canonical concepts that we developed around our own ontology). Yochai, uniquely, takes you directly to the sources and pushes you to discover and articulate your own insights about them. In addition to asking Yochai questions of life adviceor intellectual and spiritual interest, you can read the texts with Yochai's "lenses" in the margins, ask Yochai to produce lesson plans and source sheets, divrei Torah, textual outlines, and lit reviews, all backstopped by a verified library. Yochai takes its name from Shimon Bar Yochai, 2nd century sage and purported author of my namesake.

New open source releases have moved the Pareto frontier on Ramp SWE-Bench. Kimi K2.7 Code and GLM 5.2 perform stronger than their predecessors at a higher cost, within reach of closed source models.

i am so so excited about this lineup featuring some of my favorite builders for design is a breakfast party tmrw!

This is my daily driver. I am “walking the store” daily, using the chrome extension to snipe issues and fix them live. It’s a designer’s dream

Spoke at @cursor_ai’s conference, compile, yesterday hosted by @dwarkesh_sp on my research how introducing information in the latent space into LLMs can make the underlying models more efficient. Haven’t written on a chalkboard in years, was a great conference and experience 😄