Owen Sakawa, the founder of Elloe AI, wants his platform to be the immune system for AI and the antivirus for any AI agent. The idea, Sakawa explained in an interview days before the TechCrunch Disrupt conference, where Elloe AI is a Top 20 finalist in the Startup Battlefield competition, is to add a layer to company LLMs. This layer checks for bias, hallucinations, errors, compliance issues, misinformation, and unsafe outputs.
Sakawa stated that AI is evolving at a very fast pace, and it is moving this fast without guardrails, without safety nets, and without mechanisms to prevent it from ever going off the rails. Elloe AI is an API or an SDK, a module that sits on top of an AI model’s output layer. Sakawa described it as an infrastructure on top of your LLM pipeline that sits there fact-checking every single response.
The startup’s system has its own layers, which Sakawa calls anchors. The first anchor fact-checks the LLM’s response against verifiable sources. The second anchor checks if the output violates any regulations, such as the U.S. health privacy law HIPAA, the European data protection law GDPR, or if it exposes Personal Private Information. The last anchor is an audit trail that shows how all the previous decisions were made. This allows regulators or anyone auditing the system to analyze the train of thought for that model, from where it made the decision to the source of that decision and the confidence score of all those decisions.
To be clear, Sakawa said Elloe AI is not built on an LLM. In his opinion, having LLMs checking other LLMs is just putting a Band-Aid on another wound. However, Elloe AI’s system does use AI techniques, such as machine learning. There are also humans in the loop, as Elloe AI’s employees keep up with new regulations on data protection and user protection.