Pinned Tweet
At @SpaceXSI, Lauren Tan (@poteto) and Roman Ugarte (@romanugarte_) are on the core team that took Grok Bot from first line of code to launch in seven weeks.
It lets you spin up AI teammates with their own identities and logins, and they just keep working in the background. Their marketing team gave a bot its own ad budget. Lauren built an entire feature with her bot in one Slack thread without ever opening Cursor.
They joined us on the @LightconePod to talk about what it’s actually like working next to bots every day, and how the hard part is shifting from doing the work to figuring out what to do and why.
03:36 — Giving Bots Real Jobs
06:49 — One Bot or a Team of Specialists
10:39 — Sharing Bots Across the Company
14:46 — Identity and Permissions for AI Coworkers
18:37 — Bot-to-Bot Collaboration and Trust
22:35 — The Workflow Behind 4,000 PRs a Month
26:00 — Teaching Agents to Write Code You Trust
29:01 — Building a Personal Software Factory
33:44 — The Future of AI Coworkers
36:47 — Rethinking the AI-Native Company
40:13 — Teaching Agents to Check Their Own Work
43:56 — How Much Code Review Is Enough?
47:33 — Codebase Guardrails That Save Tokens
52:20 — When Product Judgment Becomes the Bottleneck
56:29 — From Finance to AI Engineering
Y Combinator retweeted
America's frontier industries need frontier financial rails.
Today we're launching Avout (YC F26).
We underwrite robotics, drone, autonomy, space, and data center companies and connect them with non-dilutive capital.
If you are building in the physical world - get in touch with us. [email protected]
avoutcapital.com
Y Combinator retweeted
Replying to @RealPact
@RealPact is a small company in the Northeast, and there are those who love it.
One of them jumped out of a plane for us.
Spenser Molloy started a real estate deal on RealPact, then went skydiving. By the time he landed, his paperwork was ready to review.
Do things that don't scale.
Y Combinator retweeted
Today we’re launching Agents by Cardinal, AI teammates built for go-to-market.
They work alongside your sales team, across every account, every day.
They know when a prospect visits pricing, when a champion changes jobs, or when a deal goes quiet. Then figure out who to talk to, why now, and what to say.
They learn how your team sells and take action across your CRM, inbox, and Slack. Or you can build your own agents from your best playbooks.
We built these agents alongside GTM teams at companies like Mintlify, Greptile, and more.
The future of GTM is here, and your team just got a whole lot bigger.
Meet your new teammates:
trycardinal.com/playbooks
Y Combinator retweeted
Your AI agent can’t search inside TikTok. Now it can.
Syncly Librarian gives Claude, ChatGPT, and Codex access to 15M+ minutes of TikTok, Reels, and Shorts, analyzed scene by scene.
Try it now at syncly.app/social/library-mc…
Y Combinator retweeted
We're launching Adside today, a swarm of agents that finds the most profitable paid strategies for you and runs them across Batch Meta, Google, LinkedIn and more.
This isn't yet another Higgsfield killer or AI ad generator. Adside agents manage the entire creative loop across every channel: they generate, import and rotate creatives, fine-tune audiences, optimize bidding and keyword targeting, and run A/B tests. Like a world-class agency would, but with agents instead of people.
Comment "SWARM" 🐝 and I'll top up your Adside account with an extra $100 in credits!
Y Combinator retweeted
We put 16 AI models in charge of buying for a produce wholesaler, every morning for 90 days. Only 3 made more money than a simple reorder formula.
The models noticed their mistakes and wrote them down, but then they mostly kept making them.
ORO AI Launch
At ORO AI, we’re training AI to run profitable businesses.
Y Combinator retweeted
Introducing Exa Websets V2
Build massive datasets from the web, now 4x faster and with 2x lower cost per result.
Also new:
- 20% higher accuracy and recall
- Pay-as-you-go pricing, no subscription
- API access for all
Y Combinator retweeted
agents keep hitting captchas and blocks on the web.
today we're launching tilion (@ycombinator): the world's fastest and stealthiest browser infra for agents.
#1 in stealth
#1 in agent action speed
#1 in start-up time
dont believe it? tell your agent:
setup tilion.com
Y Combinator retweeted
Introducing Helios: measuring 1,000+ protein biomarkers from a drop of blood to help you live younger, longer.
Here’s how it works:
• Self-collect blood at home with a painless, needle-free device
• We measure the proteins carrying signals from cells and organs throughout your body
• Your Helios AI agent explains your results, highlights risks, and offers personalized guidance
Proudly backed by Y Combinator and Formation VC.
Join the waitlist at myhelios.com
Y Combinator retweeted
Today, we're launching Clueso v2.
AI has finally become good enough to edit videos.
But creating great content still needs a human touch.
Clueso v2 is the only AI video editor where humans and AI agents collaborate in real time to create videos you actually want.
And yes, it works. We made this entire launch film in 2 days, using @clueso AI for the full edit.
Y Combinator retweeted
The average systematic review takes 67 weeks.
AnswerThis now runs one from a single research question, and shows its work on every paper it keeps or cuts.
22,760 records in, 174 studies out throughout a multistep process including screening, meta-analysis, and a PRISMA Compliant SLR at the end.
Try it here: answerthis.io/?utm_source=x&…
Run it end to end, or go step by step and overrule any call before it moves on. We hope you enjoy it!
Y Combinator retweeted
eleven years ago Karim and I were in YC S15 with an app that read your inbox, found receipts, and emailed stores asking for your money back when prices dropped. we’d applied twice. Garry was one of the people who bet on us.
Paribus could see waste. we were reading hundreds of millions of emails a day and knew exactly where people overpaid. but all we could do about it was write a polite note to customer service and hope. Ramp started from the opposite question: what if you were close enough to the money movement to actually stop it?
I didn’t expect to relearn that lesson inside our own company a decade later.
early this year, AI was breaking Ramp. a handful of people got 100x more productive and whole functions fell behind them. the teams using the least AI asked for the most headcount. it got bad enough that, while we were growing revenue 100%+, we stopped hiring for a quarter.
the problem wasn’t the models. it was the same problem Paribus had. most people could see what AI could do and couldn’t let it act. we’d tell someone in sales to use Claude Code, and the first question was “what’s terminal?” then: can it see our data? our Slack? our customers? no, no, and no.
so we connected it to everything. in Q2, 7,000 pull requests at Ramp came from non-engineers, mostly our CX team. two years ago that number was zero. a customer emails about a bug, the fix ships, and the customer gets a reply the same afternoon.
I think most of the economy is stuck in that same spot. business spend on AI tokens grew more than 5x from December to August. since then it’s gone a bit sideways, even as the models keep getting better and cheaper. model progress didn’t stall. organizations did. most companies are Paribus with AI. they can see everything and act on almost nothing.
that’s the best news I can give anyone starting a company today. the advantage isn’t access to intelligence anymore. it’s being willing to let it do the work.
thank you @garrytan, for this and for betting on two guys with a refund bot a decade ago.
Ramp started with a simple promise: help companies save time and money. Seven years later, it’s a $60 billion platform running the financial operations of tens of thousands of businesses, with more than $2 billion in annual revenue.
Now, Ramp is using AI not just inside its products, but to rethink how the company itself works.
On this episode of How to Build the Future, @tryramp co-founder and CEO @eglyman joins @garrytan to talk about the company’s journey from its earliest days and how AI has made some people at Ramp radically more productive. That shift led Ramp to rethink how work gets done across the organization, and Eric believes it could eventually let even large companies move with the speed of startups again.
00:46 — Eric’s First Startup
03:02 — How Paribus Began
05:39 — Finding the First 10,000 Users
08:17 — The Idea Behind Ramp
12:57 — Agents Need the Ability to Act
14:42 — Ramp’s Next Chapter
18:28 — Why AI Adoption Is So Uneven
20:24 — The Rise of the 100x Employee
24:28 — Inside Ramp’s AI Stack
27:07 — Getting the Rest of the Company to Catch Up
29:59 — Glass: Ramp’s Company Brain
32:32 — Teaching AI How the Company Works
36:49 — Can Big Companies Move Like Startups?
39:24 — What Cheap Intelligence Makes Possible
44:23 — When Anyone Can Ship Software
48:18 — Every Software Company Needs an AI Harness
49:45 — How Ramp Finds Great Builders
52:09 — The Power of Asking for Help
Y Combinator retweeted
Excited to officially launch @orca_aerospace today! We are building full self-driving for satellites! Orca puts an AI operator onboard each spacecraft, so satellites can make decisions and run their missions with minimal human intervention.
Y Combinator retweeted
Launching Open Actuation: High-mix, purpose-built motors and actuators.
The best actuator in the world is the one designed for your exact mission.
We enable robot and drone teams of all sizes to own every design and sourcing decision, iterate quickly, and bring truly differentiated machines to life.
Get yours at openactuation.com
Ramp started with a simple promise: help companies save time and money. Seven years later, it’s a $60 billion platform running the financial operations of tens of thousands of businesses, with more than $2 billion in annual revenue.
Now, Ramp is using AI not just inside its products, but to rethink how the company itself works.
On this episode of How to Build the Future, @tryramp co-founder and CEO @eglyman joins @garrytan to talk about the company’s journey from its earliest days and how AI has made some people at Ramp radically more productive. That shift led Ramp to rethink how work gets done across the organization, and Eric believes it could eventually let even large companies move with the speed of startups again.
00:46 — Eric’s First Startup
03:02 — How Paribus Began
05:39 — Finding the First 10,000 Users
08:17 — The Idea Behind Ramp
12:57 — Agents Need the Ability to Act
14:42 — Ramp’s Next Chapter
18:28 — Why AI Adoption Is So Uneven
20:24 — The Rise of the 100x Employee
24:28 — Inside Ramp’s AI Stack
27:07 — Getting the Rest of the Company to Catch Up
29:59 — Glass: Ramp’s Company Brain
32:32 — Teaching AI How the Company Works
36:49 — Can Big Companies Move Like Startups?
39:24 — What Cheap Intelligence Makes Possible
44:23 — When Anyone Can Ship Software
48:18 — Every Software Company Needs an AI Harness
49:45 — How Ramp Finds Great Builders
52:09 — The Power of Asking for Help
Y Combinator retweeted
We’ve raised a $7.5M seed round led by Felicis so AI can read documents as easily as it reads databases.
Finance, law and healthcare run on documents, and not databases like any other industry.
Soon AI agents will read and write more documents than humans, and they’ll need to query them the way software queries tables.
We’re building the computable layer that makes this possible.
To do this at scale, we've brought together ML researchers and engineers from Google DeepMind, Cognition, and quant trading firms.
We use computer vision and dual-stream models to turn complex multimodal documents into formats LLMs and AI agents can understand and reason over.
If your team wants to turn messy documents into reliable, AI-ready data, book a demo:
unsiloed.ai/book-demo
Y Combinator retweeted
We’re launching agent evaluations in the @LiteLLM SDK for developers who want to compare models and harnesses on their own tasks.
Before you had to write the eval loop, create temporary workspaces and collect the results in a spreadsheet.
Now you can reuse your agent, define typed tasks with a success check, and call litellm.agent_eval(). You get a scorecard with pass rate, cost, tokens and duration.





















