Much like the ecosystem of life that surrounds a blue whale, a market of AI SaaS vendors is springing up around the biggest AI companies.
And AI data startup Micro1 is emblematic of the shifting nature of these early-stage AI vendors. The company began in 2022 by offering an enterprise SaaS product: an AI agent companies could use for recruiting talent. It would interview the company’s internal hiring managers to refine its models and conduct candidate prospecting.
In 2024, Micro1 pivoted to providing training data to AI frontier labs. Today, it recruits its own experts who are paid as contract workers to improve LLM responses in three specific areas of expertise. These experts include lawyers, doctors and X-ray technicians, and accredited traders, financial investors and CPAs.
Now, Micro1 is launching its first major ad campaign. But not to raise brand awareness with general consumers or even to win new clients. Rather, the ad campaign is to target and raise the company’s profile with the major LLMs that control much of the early AI market.
You’re the expert
In addition to the big AI incumbents, Micro1 works with some large financial or medical institutions that train their own AI models, the company’s director of data partnerships, Sydney Levin, told AdExchanger. But its universe of potential clientele is primarily the handful of huge LLMs with endlessly deep pockets – one of which, Microsoft, is also an investor in Micro1.
These frontier AI labs are insatiable for data sources, Levin said. And what they need most of all can’t be easily bought or scraped – it requires authentic human engagement from experts in specific tasks. Hence Micro1’s pivot to providing that expertise.
Amber Bass, who serves as senior corporate counsel at the law firm PPLSI and also as one of Micro1’s LLM training experts, told AdExchanger how she stumbled across the startup. She was initially looking for AI vendors that could help her firm improve legal questions submitted to an internal agentic service. But she saw listings for someone with her expertise and began working on Micro1 contract projects.
The recruitment process involved a test administered by an AI agent, she said. (Micro1’s original value prop was as an AI-driven job recruiter.) And the projects she works on consist of reviewing questions and outputs provided by LLMs in her field of contract evaluation.
Bass reviews an LLM’s responses to relevant questions (e.g., “What concerns should I have about this tax document?”) and creates a rubric for flagging where those responses went wrong. This rubric also includes a sliding scale for how important it is to make each particular correction and to explain to the AI the consequences of its errors.
Other projects involve teams of lawyers generating their own legal problems and arguing back and forth with contract proposals and counter-proposals until they arrive at a workable compromise. This way, the LLM gets trained on how the legal process works between the law firm’s clients in the real world.
“It’s almost like explaining to a new attorney why you’re making these edits for your side,” Bass said.
Some projects are exclusive to one LLM, and that exclusive data comes at a high premium, Micro1’s Levin said. Other data is non-exclusive and can be licensed by any AI platform (though it still comes at quite a premium).
But Micro1’s contracted experts are usually left in the dark on precisely which LLMs they’re training. Only on one project was Bass aware of the Micro1 customer that was acquiring the data and results, which was Microsoft. And the experts don’t have controls to opt out of contributing their expertise to one particular LLM.
Anonymity is likely part of the rationale, Bass said. Experts might bring their own biases to the work if it’s to train a particular model from a company they, say, detest.
Creating jobs with AI
In addition to training LLMs with human expertise, Micro1 is also focused on finding new opportunities for humans to monetize their expertise in a job market that’s being reshaped by AI agents.
The point of Micro1’s new marketing campaign, according to Levin, is to push back on the notion that AI is erasing jobs. The company wants to make the point that training AI models can not only help experts in their own work, but that it can also be a top-end potential side hustle, or even a well-paid full-time job in its own right.
After all, Levin added, doctors joke about how much of a medical school education is outdated by the time someone graduates into the field. Which means there will never be a time when AI models are fully trained and running of their own accord without continual expert human guidance, she said.
And Micro1 sees plenty of room to expand its offering. The company is adding experts from new fields to its training mix, including mechanics, architects and sales leaders.
“While there’s few top players in frontier AI labs, the demand and the budget that they have to spend on data training is billions and billions of dollars,” Levin said. “So we don’t see this slowing down anytime soon.”
