How Spark Tank's Boldest Pitch Changes Everything in Edition 12

What we talked about.

In this edition of Spark Tank, host Sunny Ray welcomes founder Chris, an Australian military veteran building Common Layer, an AI safety layer designed to protect vulnerable users from harmful interactions with conversational AI platforms like ChatGPT and Claude. Skipping the usual slide deck, Chris explains how mental health crises, mania, and unhealthy dependency are emerging from generalist chatbots never designed for existential or therapeutic use. Drawing on his own PhD research and a book called The Silence Paradox, he outlines Common Layer's proprietary compliance frameworks, early red team testing results showing improved safety triage rates, and a dual B2C and B2B business model targeting the multibillion dollar employee assistance program market. A panel of VCs, including John Frankel, Matt Blumberg, Shane Wilson, and Lance, push back on the lack of visuals, unclear positioning between consumer and enterprise customers, and questions about defensibility against the very AI giants he is trying to protect users from. Despite the tough questioning, Chris's founder story and early traction with insurers and a police service leave the panel intrigued about where Common Layer goes next.

A veteran skips the slide deck to pitch an AI safety layer that catches what ChatGPT misses.

The questions, and the answers.

You chose to pitch without a deck today. Walk us through Common Layer and why you built it.

I founded Common Layer after seeing the rapid adoption of conversational AI causing real damage on generalist platforms being used for mental health and existential inquiry, purposes they were never designed for. Harvard Business Review confirmed existential inquiry as a top use case two years running, and without safeguards these platforms are causing suffering, from mania to depression to risk to others. We built proprietary regulatory and clinical frameworks to make conversations with AI safer.

What's the business model? How will Common Layer make money?

It's a B2C channel where consumers use a market facing version we're calling Frank, a wrapper around Claude, ChatGPT or Gemini, so it's platform agnostic. The EAP solution is a SaaS product priced around $8 a seat for larger midsize enterprise and down to $5 a seat for Fortune 500, and that's about scale and go to market.

Tell us more about your background as a veteran and health tech founder.

I'm an Aussie veteran and former startup founder who's been building health tech for ten years. I'm halfway through a PhD looking at how to make safer conversational agents and I've just written a book on that called The Silence Paradox. In my last company we built the world's first heart rate variability detection app on a smartwatch to detect stress signals for veterans and first responders, won two global innovation awards, and raised $1.5 million.

How do you prove to buyers and regulators that your layer is materially safer?

We engaged red teams to test three platforms, then two of the platforms with our layer wrapped around them. We validated that ChatGPT alone is only responding to and triaging about 35% of dangerous behaviors on its platform. When we wrapped ours around it, that jumped to 65%, and that's with only seven of the 34 variables we intend to build.

Some are calling this a tech suicide hotline via API. What do you think about that framing?

I wouldn't say anything about a suicide hotline. At the end of the day we're trying to minimize risk, we're never going to remove it entirely. OpenAI said 1.25 million people a week are displaying dangerous behaviors on their platform. We'll never totally reduce all of that, and we're certainly not going to be a suicide hotline. We do have a human in the loop solution we're building, but it's about reducing risk as much as feasible with the technology available.

What traction and market interest have you seen so far?

Market feedback has been amazing for the SaaS product going into the EAP market, worth about 250 billion dollars across the US, Australia and the UK. We've got letters of intent underway with the Queensland Police Service, Zurich Insurance, and Benoir Insurance, a former investor in my last business. We've also spoken to three large EAP vendors who are very interested in prototypes and pilots.