Sumble emerges from stealth with $38.5 million to bring AI-powered context to sales intelligence

Ask any salesperson how much information they want about a potential customer and you’ll never hear the end of it. This is the foundation that drives the crowded sales intelligence market, which today has services that can do everything from helping identify and background prospects, to writing the pitch and conducting independent follow-ups.

But sales teams need more than just data; They want context. Sumble, a startup based in San Francisco, attempts to provide this context by scouring the web across social media, job boards, company websites, regulatory filings, etc., to uncover information about what’s happening inside companies.

The brainchild of Anthony Goldblum and Ben Hamner, founders of data science and machine learning community Kaggle, Sumble uses a knowledge graph powered by large language models to connect the various data points it collects. The result, Goldbloom told TechCrunch, is a comprehensive view of a company’s technical data — what tools are being used in which departments, which projects are being launched or running, its organizational chart, what technology the company might be looking to adopt, and, most importantly, who to contact.

But given how crowded this market already is, from established companies to countless AI sales development agents, the question is: Does the world really need more?

Goldbloom says the startup’s approach appears to be working: He told TechCrunch that since launching in April 2024, the startup has signed on 17 enterprise customers, including Snowflake, Figma, Wiz, Vercel, and Elastic, and has tens of thousands of users in total. About 30% of its users pay for a Pro subscription, either themselves or their company, and so far, growth has been driven by word of mouth. The startup declined to share details of its revenue, but we know that revenue increased by 550% year-on-year.

“What usually happens is we spread quickly within the company,” Goldblum said. “We will go from 1 to 500 MAU [monthly active users] In the company for six months. The way diffusion happens is usually within a Slack channel, then within one team, then within an office, then within that company.

Attraction, quality of customers and strong customer retention played a big role in attracting investors’ attention, Goldblum said. The startup came out of hiding on Wednesday with $38.5 million in funding — Kuato led an $8.5 million seed funding round, while Canaan Partners led a $30 million Series A funding round. AIX Ventures, Square Peg, Bloomberg Beta, Zetta and investors have also invested Angels, including Salesforce CEO Marc Benioff and former GitHub CEO Nat Friedman.

Notably, Sumble’s founders have attracted investors they know inside out. Rich Boyle, currently a general partner at Canaan, was an observer to Kaggle’s board, while Bloomberg Beta and Zetta were also at Kaggle’s table. Goldbloom co-founded AIX Ventures, though he told TechCrunch that he left the company when it considered investing in Sumble.

However, Semple faces a lot of competition. Competitors include Apollo.io, Slintel, SalesLoft, Cognism, Repup.io, ZoomInfo, HubSpot, Outreach, and others that offer more focused point solutions or comprehensive sales IT toolsets. Since Sumble currently uses publicly available data, there’s nothing to stop others from doing what you’re currently doing.

However, Goldbloom is confident that Sumble’s moat is deeper than it first appears, thanks to how it structures its knowledge graph, which covers about 2.6 million companies around the world.

“The way we think about it is that the more data we add to the knowledge graph, the richer the text becomes. We view the richness of the knowledge graph as a tremendous source of defensibility,” he said.

Sumble is also counting on the continued adoption of large language models to help it continue to scale as it expects to see people using AI alongside its service. “The way we structure our data is that the knowledge graph will always be queryable by large language models […] “The idea is that you can ask ChatGPT for Apple’s text stack, or you can ask ChatGPT for Apple’s technical stack based on our data,” Goldblum said.

“We believe that AI will change the data vendor landscape so much that having a knowledge graph architecture as a way to feed context into the large language model will be a key part of the LLM ecosystem,” he added.

The service is currently offered as a web app and via an Application Programming Interface (API), and there is also a paid plan that offers more features such as workflow and customer relationship management (CRM) integrations, as well as notifications when development on a lead is of interest.