Felix, officially known as Spoken Empathy Systems Inc., launched on Tuesday with $1.7 million in pre-seed funding to expand its platform that automates complex professional workflows without constant human oversight.

XYZ Venture Capital led the round with participation from angel investors, including current and former leaders and founders at Amazon Inc., Apple Corp., Palantir Technologies Inc., FlexPort Inc., Yelp Inc. and Midjourney Inc.

Felix is already deployed by major legal, finance and insurance firms, who use it to automate institutional work in regulated environments. The company said it will use the funding to expand its product capability and grow.

Current foundation artificial intelligence models are probabilistic, so inputting the same prompt more than once can get different answers. This can be problematic for industries where accountability and reliability are a necessity. Felix takes workflows and turns them into reliable steps where automations generate the same outputs every time, with complete traceability — a must for regulated industries. 

“Most AI tools help you do the same work faster,” co-founder and Chief Technology Officer Mato Vetrak said. “We’re building the kind of safe, scalable and auditable infrastructure that high-stakes professional services can trust to codify who has the authority to, and how to make judgments about, different use-cases, such as processing mortgage applications.”

Felix operates as a sort of routing system that determines when AI is needed and applies its reasoning only when needed. It codifies all processes and maintains a full audit trail. The process builds all the automation code and runs it continuously; this way, all automations are themselves deterministic and run the same every time.

The platform only routes through large language models when interpretation is required. This routing model also saves on AI token costs in high-volume operations and increases overall reliability, the company said.

One United Kingdom-based company, Recourse Collective Ltd., an insolvency firm, built a fraud detection engine atop Felix to analyze creditor claims at scale. According to founder Matthew Letts, the platform allowed the company to cross-reference tens of thousands of claims documents for “phoenixing” fraud.

That’s a type of fraudulent activity where a company dissolves and then resurrects itself, much like the legendary bird, to escape debt. Within eight weeks, the system identified £1.4 billion, about $1.8 billion, in recoverable assets.

“Before we started working with Felix, we thought we might find millions in recoverable assets,” Letts said. “But the scale that Felix gave us to automate big, thorny, repetitive tasks uncovered far more than we expected.”

Image: SiliconANGLE/Microsoft Designer