Invisible Technologies CEO On Building AI Around Real Workflows, Not Hype

What does it actually take to make AI work inside a real business, where messy data, human judgment, and operational risk all collide?

In this episode, I sit down with Matt Fitzpatrick, CEO of Invisible Technologies, to talk about why the biggest barrier to enterprise AI is not model quality, it is everything that comes before the model ever gets to work.

Since stepping into the CEO role in January 2025, Matt has moved quickly, raising $100 million and expanding Invisible's footprint across major cities including New York, San Francisco, DC, Austin, London, and Poland. But this conversation is far less about headlines and far more about what happens in the trenches of AI adoption, where companies are trying to move from pilots and PowerPoint promises to systems that actually deliver results.

A huge theme throughout our discussion is data readiness. Matt makes a compelling case that most businesses are still dealing with fragmented systems, inconsistent records, and information spread across disconnected tools. That reality makes it incredibly hard to deploy AI in a way that creates trust and value.

We talk about SwissGear, where Invisible used its Neuron platform to clean and structure 750 scattered tables in just one week, a task that could have taken a large engineering team months or longer. We also discuss why that kind of work matters so much, because once the data foundation is fixed, companies can start making better decisions on forecasting, operations, and planning with a level of confidence that simply was not there before.

We also spend time on Invisible's human-in-the-loop approach, which I think will resonate with a lot of listeners trying to cut through the noise around job displacement and agentic AI. Matt argues that the real opportunity is not replacing people, but giving them better tools to handle repetitive work while preserving room for human expertise, judgment, and oversight.

He shares examples from commercial credit workflows, healthcare, and sports analytics, including a fascinating story about the Charlotte Hornets using AI to turn broadcast footage into detailed tracking data. What stood out to me was how practical his perspective felt.

This was not theory. It was about building systems around how organizations actually work, rather than expecting businesses to reshape themselves around a generic AI product.

Another part of the conversation that deserves attention is governance. As boards rush to understand agentic AI, Matt explains why trust, standards, and responsible deployment are now driving buying decisions just as much as raw capability.

We talk about privacy in healthcare, the risks of scaling autonomous systems without mature governance, and why enterprise adoption still trails consumer AI by a wide margin. That gap between excitement and execution may be one of the most important stories in AI right now.

If you are wondering why so many AI projects never make it into production, or what it will take for enterprise AI to finally deliver on its promise, this episode is packed with insight. It is a conversation about data, deployment, governance, and the role humans will continue to play as AI becomes part of everyday business operations.

After listening, I would love to know where you stand, is the future of AI really about bigger models, or is it about making AI fit the messy reality of how work gets done?