- Premium Content It takes our newsroom weeks - if not months - to investigate and produce stories for our premium content. You can’t find them anywhere else.
Glenn Kaonang · · 5 min readWhy Datasaur is chasing private AI over scraps in data labeling
Meta’s blockbuster investment in Scale AI in June rattled the data labeling industry. Scale’s clients scrambled to assess their partnerships with the company, wary of handing over proprietary datasets to a firm now plugged into Meta’s empire.
Some of this business flowed to Ivan Lee, CEO of US-based data labeling startup Datasaur.
“We’re talking to clients that clearly were previously clients of Scale and are now looking to move away from Scale because they no longer trust them,” Lee tells Tech in Asia. “It’s a bit of a free-for-all in this space right now.”
Datasaur CEO Ivan Lee / Photo credit: Datasaur
Datasaur, which tags text and audio at scale, was founded in 2019, years before the AI wave went mainstream. Before launching the firm, Lee had worked at Apple managing training data for machine learning projects but found existing labeling tools inefficient.
Still, Datasaur, which has raised about US$9 million from investors, including OpenAI president Greg Brockman, doesn’t want to just pick through Scale’s leftovers. Instead, it has recently switched to deploying private large language models (LLM), which are run within a company’s own environment rather than shared publicly.
The company claims that private LLM deployments now make up 50% of its annualized revenue, a shift that could push it to profitability by the end of the year.
Why private LLMs?
Lee points out that most companies can’t afford the US$10 million price tag for OpenAI’s customized AI consulting services, which involve fine-tuning models with clients’ proprietary data. For Lee, that’s OpenAI’s way of saying its own offering, ChatGPT Enterprise, isn’t “good enough.”
Datasaur is targeting this middle market: companies that need custom, private AI solutions but can’t justify Big Tech’s hefty asking price.
OpenAI CEO Sam Altman / Photo credit: Shutterstock
The pivot to LLMs fits its roots. The company made its name in labeling for natural language processing, with clients ranging from major US banks to the FBI and AI Singapore. Now, the same skills – understanding how machines parse language, structure information, and draw inferences – can be used to deploy and refine LLMs.
Datasaur now helps enterprises clean and organize their data before feeding it into AI systems.
See also: From search to super agent: Genspark takes big AI leap
Crowded field
No clear playbook
Stay ahead in Asia’s tech landscape
This is premium content. Subscribe to read the full story.
Datasaur pivots from AI text labeling to deploying private language models, reflecting demand for cheaper, more tailored AI beyond those offered by Big Tech.
We know this is not ideal. ⌛ Sign up in 20 seconds. Cancel anytime.
Our subscriber community includes professionals from these companies:
Stay updated on the go with our mobile app.
Get latest insights with smoother, more personalized experience through TIA mobile app.
