Welcome to Inc.’s 1 Smart Business Story! Today we are looking at what it means to build a fast-growing company with AI at the core–not just as a buzzword, but as a real growth engine.
OpenAI is increasingly turning some of its biggest startup customers into poster children of the AI revolution. Its new Frontier Builders campaign spotlights founders from companies like Decagon, Valthos, and Clay that are using OpenAI’s large language models to power real, revenue-driving products like Decagon, Valthos, and Clay.
One of those standouts is Clay, a go-to market platform that uses OpenAI’s models to automate and supercharge sales outreach. The company recently crossed $100 million in annual recurring revenue, scaling from $1 million in ARR just two years prior.
In this piece, Inc. reporter Ben Sherry speaks to Clay founder Kareem Amin and OpenAI’s startups head Marc Manara about:
How Clay turned AI agents into a powerful customer-acquisition engine
What it looks like to build and scale an AI-first business
Why OpenAI is showcasing startups like Clay as examples of the next wave of company building
Tell me how you’re seeing AI-first startups redefine their industries at [email protected].
Sales Startup Clay Just Hit $100 Million in ARR
Leveraging AI, Clay CEO Kareem Amin turned GPT-powered agents into a ‘customer acquisition machine.’
BY BEN SHERRY, STAFF REPORTER
Kareem Amin isn’t a big pottery guy, but he’s shaping Clay, his startup that helps go-to-market and sales teams find and pursue business leads, into an industry powerhouse. That’s due in large part to the company’s prodigious use of OpenAI’s large language model technology.
Amin is one of several founders currently featured in a new OpenAI brand campaign that spotlights “frontier builders” who are “redefining entire categories of business,” with AI at the core, according to OpenAI head of startups Marc Manara. These companies are already successful, says Manara, but they’re poised to reap even greater benefits if OpenAI’s tech continues to exponentially improve.
Manara leads a global team of nearly 50 people who help some of OpenAI’s startup customers push the limits of what large language models can do in specific industries. Through shared Slack channels and weekly meetings, Manara’s team keeps customers like Clay informed of changes to OpenAI’s API, which enables companies to integrate LLMs into their own products. The team also assists in customizing the API to suit customers’ specific needs, and helps troubleshoot prompts to improve the models’ output.
On January 22, CEO Sam Altman posted on X that the API business had added over $1 billion in ARR over the past month.
Clay uses OpenAI’s API to power agents that can automate much of the busywork that goes into sales prospecting and generating leads for business. When Amin cofounded the company in 2017, he wanted to “help people turn any of their ideas into programming,” and make software development as intuitive as molding clay. Amin says his vision was similar to the current vibe coding revolution, but the technology just wasn’t ready in the 2010s, so after a few years he pivoted to focus on helping go-to-market and sales pros to be more creative and intentional with their campaigns.
Imagine you want to identify potential customers for your business selling AI-powered clown shoes; you could upload a list of circuses, and then ask Clay’s OpenAI-powered agent (called Claygent) to enrich that list by searching the web for the each circus’ point of contact and obtaining their email address and phone number. You could even ask Clay to surface negative reviews that might help to strengthen your pitch. Once you have enough actionable prospecting intelligence, you can use Clay to craft your actual outreach message and customize it for each potential client.
This agentic system has catapulted Clay to massive success; in December the company announced that it had crossed $100 million in annual recurring revenue, growing from $1 million in ARR just two years prior. In August, the company raised $100 million at a $3.1 billion valuation. Amin says that Clay’s popularity has turned the company into one of OpenAI’s biggest API customers, with the ChatGPT-maker’s models processing and generating trillions of tokens on Clay’s behalf.
OpenAI isn’t the only model provider that Clay relies on, though. The company also uses Anthropic’s and Google’s APIs for access to their models, Claude and Gemini respectively.
Businesses pay for the API based on the number of tokens processed and generated by AI models. Generally, Manara says, when OpenAI detects that a startup’s token usage is increasing, they’ll reach out and offer assistance, but they also place bets with early-stage teams that have recently raised seed funding and have potential. OpenAI also uses these relationships to get feedback that helps to improve future models at completing economically viable work, a key part of their quest to build artificial general intelligence.
Increasingly, Manara’s startup team is working with venture capital firms to “capture the signal” of companies that could be the next frontier builders, a process that also heavily involves Clay. Manara says he and his team use the platform to gather and enhance data about funding for new AI-driven companies from both third-party sources like Crunchbase and Pitchbook, and the first-party information it obtains from its VC partners. “We’re a really small team, we can’t throw people at the problem,” Manara says, “but Clay can help us be really precise and targeted when we do outreach or we’re trying to make sense of the inbound that’s coming to us.”
Amin says that Clay even helps OpenAI to identify potential prospects based on their ChatGPT usage. If Clay finds OpenAI data that suggests a few high-level employees from the same company have recently started making heavy use of ChatGPT, it could recommend reaching out to their manager and pitching them on an enterprise license.
“What if we knew who your best customer is, and then could predict what properties would make this other person the next-best customer,” Amin says of his work with OpenAI. “What if we could project the next frontier customer on OpenAI, using all the information that we have.” In practice, Amin refers to his platform as a “customer acquisition machine.”
This year, Amin says that Clay will release AI-powered agents that have been specialized to excel at industry-specific tasks. The first of these agents will be focused on the financial sector, and another realtor-focused one will soon follow.