Liftoff with Keith Newman: For Founders looking for a new growth playbook and further inspiration. · Former Journalist, Startup + GTM exec Keith Newman

Trustworthy AI: Why Explainability Will Define the Next Decade of Enterprise Technology

·37 min
Keith opens by welcoming Pat Kondo, whose company sits inside Virginia’s fast-growing data-center world. Pat lays out the scale of that ecosystem, with engineers, construction crews, electricians, plumbers, shipping, and a whole support economy building around it. The conversation treats that as a sign of how much infrastructure is going into AI work. From there, Keith turns to explainability and why it matters. Pat says it is not a philosophical extra; it is a practical requirement when systems shape decisions people rely on. He frames it as a sliding scale. At one end are lower-stakes use cases like pulling financial information from documents. At the other are critical systems where a bad answer could injure people, cause death, or create a serious catastrophe. Keith presses on whether that logic applies outside mission-critical settings, and Pat says yes. Even ordinary enterprise workflows can end in lawsuits if the model is wrong. The conversation then shifts to defense technology, where explainability becomes non-negotiable. Pat points to the scale of investment flowing into defense tech and cites the Golden Dome missile defense system as an example of the level of explainability required. Keith then asks about Seeker itself, and Pat says the company is about 60 percent commercial and 40 percent government. He expects that mix to flip next year. That shift, he says, comes from rapid demand across government departments that want better efficiency and AI systems for revenue leakage detection. Pat gives a simple example: organizations often have legacy accounting structures that hide the real number of accounts, and once those accounts are consolidated and reviewed properly, savings show up quickly. The discussion keeps coming back to one point: good AI depends on good data. Pat closes by comparing it to earlier work building inertial measurement systems for spacecraft. The principle is unchanged, he says, because bad data still sends you to the wrong place. Keith leans into that point and ends on the idea that the right path in AI still starts with trustworthy inputs and accountable outputs.

As heard by us

A practical AI conversation about explainability as a buying requirement.

Pat Kondo treats explainability as a working requirement, not a slogan, and that gives the conversation some weight. The discussion moves from Virginia's data-center buildout to the mechanics of explaining why a system made a choice, then to the costs of getting it wrong when…

Read the full review in PlayNext →

Why you'd press play

You want the real-world case for explainable AI, not buzzwords.

Read the full recommendation in PlayNext →
Listen to the show on