Owen Sakawa, founder of Elloe AI, wants his platform to be the “immune system for AI” and the “antivirus for all AI agents.”
The idea, Sakawa said in an interview a few days ago at the TechCrunch Disrupt conference, where Elloe AI was named a Top 20 finalist in the Startup Battlefield competition, is to add a layer to a company’s LLM that checks for bias, hallucinations, errors, compliance issues, misinformation, and unsafe output.
“AI is evolving at such a rapid pace that there are no guardrails, no safety nets, no mechanisms to keep it from going off the rails,” said Sakawa (pictured above).
Elloe AI is an API or SDK, a module that sits on top of the output layer of an AI model, or, as Sakawa described it, “the infrastructure that sits on top of the LLM pipeline.” “And it’s put there to basically fact-check every answer.”
The startup’s system has its own layer, or “anchor,” to use Sakawa’s words.
The first anchor fact-checks the LLM response against verifiable sources. The second anchor then checks whether the output violates regulations such as the U.S. medical privacy law HIPAA, Europe’s broader data protection and privacy law GDPR, or if some personally identifiable information (PII) is exposed. The final anchor, Sakawa said, is an audit trail that shows how every decision has been made so far, allowing regulators and those who audit the system to “analyze the thought flow of that model, the thought flow of the model on which that decision was based, and the confidence score of every decision.”
To be clear, Sakawa said Elloe AI is not built on LLM. Because, in his opinion, LLMs checking other LLMs is simply putting a “band-aid on another wound.” However, Elloe AI’s systems use AI techniques such as machine learning. And humans are also involved in this information. Elloe AI’s employees are constantly adapting to new regulations regarding data protection and user protection, Sakawa said.
tech crunch event
san francisco
|
October 27-29, 2025
If you want to learn more about Elloe AI, check out dozens of other companies, hear their pitches, and hear from guest speakers on four different stages, join us at Disrupt in San Francisco from October 27th to 29th. Click here for more information.
