The founders who built Kaggle turned the open web into a knowledge graph - and pointed it straight at the least glamorous job in software: figuring out what a company actually does.
Here is a secret about sales that nobody puts on a billboard: most of it is research. Before a single email goes out, someone has to figure out what a target company uses, what it is building, and who inside it can actually sign a check. It is unglamorous, it is slow, and it is the part everyone skips. Sumble's bet is that this is precisely the part worth automating.
Sumble is a San Francisco startup that crawls the public web - job boards, company websites, social media, regulatory filings - and feeds all of it into a knowledge graph underpinned by large language models. The output is not another contact list. It is context: which tools a company uses in which department, what projects it has underway, what technology it is looking to adopt, its org chart, and, crucially, who to talk to. The graph currently covers roughly 2.6 million companies worldwide, which is a lot of companies for a team of about 29 people to have mapped.
The leverage, of course, is the point. A company this size covering a dataset this large tells you the machines are doing the reading. That is the whole trick, and it is a trick these particular founders have pulled before.
Anthony Goldbloom and Ben Hamner are not first-timers. They built Kaggle - the data-science and machine-learning community that became a fixture of the field - and sold it to Google in 2017. Goldbloom ran it; Hamner was the technical backbone. They spent years watching how raw data becomes a decision, which is a useful thing to have watched if your next company is, essentially, a machine for turning raw web data into sales decisions.
They founded Sumble in 2022 and then did something increasingly rare: they stayed quiet. The company operated in stealth for roughly three years. It launched its product publicly in April 2024 and only emerged from stealth, with a press release and a funding announcement, in October 2025. By then it had already signed 19 enterprise customers. The order matters. Build first, announce later.
The incumbents in this market - ZoomInfo, Apollo, Cognism and the rest - largely sell you a database of people: names, titles, emails, phone numbers. Useful, up to a point. Sumble's argument is that a phone number is worthless if you do not know why you are calling. Knowing that a company just posted twelve jobs for a data platform, or that it runs a particular stack in a particular department, is the difference between a pitch that lands and one that gets deleted. Whether that argument holds up against entrenched competitors is the open question. The early numbers suggest it might.
Consider the conversion rate. Roughly 30% of Sumble's users move from free to paid Pro plans. In freemium software, where the standard outcome is that almost nobody pays, 30% is the kind of figure that makes investors sit up. And the growth is not driven by a sales army - it spreads the way good tools spread, one forwarded link at a time, from a single user to hundreds inside the same organization. The company describes going from one to 500 monthly active users inside a single account in six months. That is product-led growth working as advertised, which is not something you get to say often.
Built Kaggle into the definitive data-science community and sold it to Google in 2017. Now runs Sumble, applying a decade of watching data become decisions to the go-to-market problem. Based in San Francisco.
Kaggle's technical backbone and co-founder. At Sumble he leads the engineering behind a knowledge graph that maps millions of companies with a small team and a lot of language-model horsepower.
Sumble ships as a web app and an API, with paid plans adding CRM integrations and prospect-activity alerts. Here is what that buys a go-to-market team.
Which tools a company uses, broken down by department - not a guess, but signals pulled from the public web.
Org charts and reporting lines so reps know who actually owns the decision before the first call.
Active projects and initiatives a company has underway - the timing signals that make outreach relevant.
What technology a company is looking to adopt, inferred from hiring patterns and public activity.
Who to reach, connected to the context of why - the last mile that turns research into a conversation.
API access, CRM integrations, and activity notifications for lead enrichment and targeting at scale.
Sumble disclosed two rounds together when it left stealth in October 2025: an $8.5M seed and a $30M Series A. The cap table reads like an enterprise-software guest list.
Coatue led the seed. Canaan Partners led the Series A, with participation from AIX Ventures, Square Peg, Bloomberg Beta, and Zetta. The angel list is the eye-catching part: Salesforce CEO Marc Benioff and former GitHub CEO Nat Friedman both put money in. When the person who runs the largest CRM company on earth backs a sales-intelligence startup, it is worth noticing.
Goldbloom and Hamner sell the data-science community they built to Google.
The pair reunite in San Francisco to start Sumble, working in stealth.
Sumble opens its account-intelligence web app and begins signing enterprise customers.
Seed and Series A announced together; 19 enterprise customers and 550% YoY growth disclosed.
"Context beats contact info. Sumble reads the web so your reps don't have to guess what a company is building."
Go-to-market teams at companies of all sizes, with tens of thousands of total users. Named enterprise customers at launch included Snowflake, Figma, Wiz, Vercel, and Elastic.
Sumble builds an AI-powered knowledge graph of company data - crawling job boards, company sites, social media, and filings - to give sales and go-to-market teams technographics, org charts, active projects, and contacts for target accounts.
Anthony Goldbloom (CEO) and Ben Hamner (CTO), the founders of the data-science community Kaggle, which they sold to Google in 2017. They founded Sumble in 2022.
$38.5 million total: an $8.5M seed round led by Coatue and a $30M Series A led by Canaan Partners, announced in October 2025, with angels including Marc Benioff and Nat Friedman.
Go-to-market teams at companies of all sizes. At its 2025 launch it had 19 enterprise customers including Snowflake, Figma, Wiz, Vercel, and Elastic, plus tens of thousands of total users.
Rather than focusing mainly on contact databases, Sumble emphasizes context - a knowledge graph of what companies use, build, and need - across roughly 2.6 million companies, assembled largely by large language models reading the public web.