A VP of Engineering tweets about scaling problems at 11:42 PM on a Tuesday. By Wednesday morning, three sales reps are already in her inbox, each pitching infrastructure solutions she hasn't publicly said she's shopping for. Call it prescient, call it invasive—either way, it's increasingly how enterprise software gets sold.
Autumn AI, which emerged from Y Combinator's Winter 2026 cohort, wants to make that kind of hyper-responsive outreach the norm rather than the exception. The San Francisco startup's thesis is deceptively simple: in B2B sales, timing matters more than your CRM thinks it does. A perfectly crafted pitch delivered two weeks late is just noise. Get there first, when the problem is fresh and the budget hasn't been allocated, and you might actually have a conversation worth having.
"We're building an AI maximalist sales platform," says Vishnu Sampathkumar, the company's co-founder and CEO, a phrase that manages to sound both ambitious and vague in equal measure. What he means, in practice, is this: monitor everything prospects do online, figure out what signals intent to buy, then tell your sales team before anyone else does.
The product, which launched in January, scrapes public data sources most sales teams probably don't have time to watch—LinkedIn updates, GitHub commits, company blog posts, incorporation filings, conference speaker lists. Sales teams configure the platform to flag specific activities: maybe it's a VP posting about infrastructure headaches, or a stealth-mode founder filing paperwork in Delaware. When something matches, Autumn AI fires an alert into Slack or email.
Chasing Founders Before They Have Websites
The most revealing example of how this works comes from a February LinkedIn post—unverified by outside sources, but specific enough to be interesting. According to Autumn AI, Brex used the platform to identify more than 150 venture-backed founders in a single week, before those companies had been publicly announced.
That's either impressive or slightly unnerving, depending on where you sit. For sales teams selling corporate cards or cloud infrastructure to startups, reaching founders before TechCrunch does is the whole game. For founders who haven't finished incorporation paperwork yet, it means the pitches start arriving faster than they probably expected.
Brex's logo appears on Autumn AI's website alongside Corgi and Mintlify, though the company hasn't independently confirmed these customer relationships. (A spokesperson for Autumn AI did not respond to requests for comment.)
Sampathkumar and CTO Shiv Kampani both come from technical backgrounds—Columbia connections, machine learning roles, the usual pedigree for a B2B SaaS startup with AI in the pitch deck. Sampathkumar previously sourced tech investments at Sierra Ventures and worked as an ML engineer at Modulus, an AI security firm. Kampani has AI research credits and presented at NeurIPS after working with SandboxAQ, the quantum security spinout from Alphabet.
LinkedIn suggests the company now employs somewhere between two and ten people, which puts it squarely in the "still figuring things out" stage of startup life. The pitch, though, leans heavily on long-context language models—the idea being that Autumn AI doesn't just hoover up signals, but synthesizes them into something resembling genuine buyer intent.
Whether that distinction holds up in practice is another question entirely.
A Market Already Drowning in Signals
Here's the uncomfortable reality for Autumn AI: the sales intelligence category is both crowded and cynical. ZoomInfo rolled out expanded real-time buyer signals back in October 2024. 6sense launched something called "Signalverse" in November 2025—yes, they actually called it that—claiming to capture trillions of buyer signals through an AI-powered orchestration layer. Seismic embedded AI insights into its winter product release in February.
Intent data, as a category, has a credibility problem. A blog post from Lift AI has argued that many intent signals clock in below 20 percent accuracy. The math is brutal: if you're flooding sales teams with alerts that turn out to be false positives four times out of five, you're not delivering intelligence. You're delivering spam with a dashboard.
Autumn AI's bet appears to be that newer models can cut through that noise more effectively. Maybe they can, maybe they can't. The startup is too young to have a meaningful track record either way.
The platform does offer practical integrations—API access, webhooks, CSV exports, direct S3 bucket pushes—so teams can route intent data into existing workflows without ripping out their entire tech stack. That's table stakes in 2026, but it matters. Sales leaders are exhausted by tools that promise to change everything and then require six months of implementation before anyone can log in.
The Challenger's Puzzle
As of late March, Autumn AI hasn't announced funding beyond Y Combinator's standard investment. Its privacy policy was updated March 2, and its GitHub organization remains empty—either because the team is still building in stealth or because there's nothing to open-source yet.
For sales leaders evaluating the platform, the value proposition is straightforward enough: get alerts when prospects do things that suggest they might be ready to buy, ideally before your competitors notice. The emphasis on monitoring diverse sources—not just website visits or white paper downloads, but social posts and public filings—positions Autumn AI as a tool for outbound-first teams that need speed over nuance.
The harder question is whether "AI maximalist" translates into sustained competitive advantage. In a market where incumbents already tout trillions of signals and agentic orchestration (whatever that means this quarter), Autumn AI faces the classic early-stage dilemma: prove that newer is materially better, or risk becoming just another logo on the slide deck explaining why intent data doesn't work as well as vendors promised.
Signal volume is easy to scale. Signal quality—the kind that doesn't make your sales team stop trusting the alerts after two weeks—remains elusive. Autumn AI has a few quarters, maybe a year, to demonstrate it's cracked that problem in ways ZoomInfo and 6sense haven't.
Until then, it's another startup betting that better AI can solve sales' oldest frustration: showing up at the right moment, when the prospect is actually ready to listen.
