ben hylak

@benhylak • 50,631 subscribers

we are hiring. cto @raindrop_ai, prev @apple design, @spacex eng

Shorts

753,951 views

34,857 views

Videos

sweetdream.ai

Watch Anya Live

Anya is streaming live right now! Join her private show and enjoy exclusive content.

Private Show

Join now for exclusive access

Free preview available • Premium content

63,450 views • 17 hours ago

191,803 views • 21 days ago

197,581 views • 1 month ago

313,502 views • 2 months ago

417,729 views • 6 months ago

179,508 views • 3 months ago

27,972 views • 14 days ago

891,471 views • 1 year ago

180,888 views • 4 months ago

151,529 views • 7 months ago

62,693 views • 3 months ago

160,824 views • 11 months ago

49,417 views • 3 months ago

140,381 views • 1 year ago

77,360 views • 10 months ago

96,020 views • 1 year ago

117,117 views • 2 years ago

31,575 views • 5 months ago

46,207 views • 1 year ago

54,561 views • 2 years ago

No more content to load

Live Cam

ben hylak

Shorts

make music physical again

vibe coded our new website last night

Videos

Watch Anya Live

today, we’re excited to announce raindrop 2.0: self-healing agents. we now train custom models that autonomously detect hidden issues with your agent. i could tell you all about it, but wouldn’t you rather hear it from someone… else?

we built the first sane way to debug your agent locally. you can see your traces. codex/claude code can too. this lets them write evals and test your agents automatically. best part: it's completely free and open source. install with 1 line. (github below)

now, your agent can fix itself. introducing raindrop triage. an agent for finding and investigating agent issues.

on saturday, i vibe coded a little app for collecting + virtually polishing rocks. i just realized today that it works on any object. behold, a polished can of lacroix.

earlier this year, we launched the first agent monitoring platform now, we are used by some of the fastest growing ai companies (@replit, Framer, Speak, Clay) and just raised a $15m seed round

we’re excited to announce trajectory explorer: the first sane way to navigate agent traces. every decision your agent made is now searchable in seconds only in Raindrop

.Speak is an AI language learning app that serves over 15 million users. cto/co-founder Andrew Hsu discusses how speak approaches agent engineering, the importance of self-healing loops, and how they use Raindrop

just tried it and... it's really, really bad. leaves me wondering if the demo was fake?

yesterday, i built the code review app i've been begging twitter for. - live streams change summaries as claude code works - quizzes me about changes before i commit them. - finds bugs locally -- without having to push to GitHub

today, we're announcing our fundrai-- jk. we're shipping. meet raindrop experiments 🧪

today, we're introducing self diagnostics: the first ever way for agents to proactively self-report issues they encounter. welcome to the future of agent observability.

Today, we're excited to announce Deep Search. It’s like Deep Research, but for your production AI data. Search for anything, and Raindrop automatically trains little models to accurately classify any topic or issue, across millions of events.

who said b2b saas has to be ugly? new in Raindrop: conversation x-rays🩻

AI products fail constantly—in ways both hilarious and terrifying. Regular software throws exceptions. But AI products fail silently. Meet Raindrop : the first Sentry-like monitoring platform for AI products.

AI companies today are faced with a choice: build great AI products, or protect user privacy. That compromise ends today. Introducing Raindrop’s PII Guard.

today we're launching Dawn -- the first product analytics platform built for ai products. with daily slack digests, automatic topic detection, powerful filtering, and semantic redaction rules, we're making it easier than ever to keep a pulse on your ai product.

in the future, ai will improve the products you use, before you even have a chance to leave feedback. to that end, i’m excited to announce Dawn, which i’ve been building with my best friends zubin koticha and alexis

at Raindrop, we like to treat every prompt as a function: structured inputs, structured outputs every function used to prepare the input for the model lives in the same prompt file, and we have an extension for iterating on the prompt right inside our codebase.

here's the talk I gave at yesterday: designing unbounded products lessons from vision os, good/bad ai product examples, thoughts on the future, and more below 👇

implementing production ready ui with figma + gpt-4 vision 🤯