That line alone made this one of my favorite conversations we’ve had on Not Another AI Podcast so far.

In this episode, Paras Doshi, Head of Data at Opendoor, breaks down what it actually takes to move an organization toward AI-native decision-making with Shanti Greene and me.

Among various other things, we talked about:

* Why most companies still struggle with trust in AI analytics

* How verification loops became necessary after a rollout

* Why AI increased a data team’s workload instead of reducing it

* The shift from dashboards to conversational analytics

* Why “deterministic enough” may be the future of enterprise AI

If you work in AI, analytics, data engineering, or enterprise tech, this episode will feel very real.

In this episode, Piyanka, Shanti, and Ant dig into why that skepticism exists and what builders and buyers can actually do about it.

The conversation covers the gap between demos and production, why confident AI outputs can be more dangerous than uncertain ones, and what real due diligence looks like before signing a contract. There is also a candid look at vibe coding, black-box development, and the security vulnerabilities that tend to show up nine months after you thought the product was done.

The episode ends where the harder questions live: AI bias, explainability, and whether the industry is asking the right things about the training data underneath all of it.

In this episode, Piyanka, Shanti, and Andrew unpack what you are really paying for with AI. The conversation moves from subscription tiers and tool sprawl to a harder question. Where does actual value come from?

They talk about why most AI products struggle to justify their pricing, why effort-based models are breaking, and why outcome-based or gain-sharing models may take their place. There is also a candid look at how easy it is to mistake a polished demo for a real product.

The episode ends on a more important shift: AI that stops at insights is not enough. The real value shows up when it drives action.

In this episode of Not Another AI Podcast, Piyanka, Shanti, and Andrew unpack the idea of agency in AI, how it differs from using LLMs, and why outcomes matter more than the technology itself. They also tackle where companies should start, when agents actually make sense, and why most organizations chase AI before fixing their data foundations.