What happened
Arsalan Tavakoli-Shiraji, co-founder of Databricks, is scheduled to speak at TechCrunch Disrupt 2026 this October in San Francisco. According to a preview from TechCrunch AI, his session will focus on why enterprise AI deals are stalling. The core argument is that large organisations have shifted their evaluation criteria. They are no longer buying based on exciting pilots, but on whether a product is safe and operationally stable enough for broad deployment.
How the room's reading it
The consensus among enterprise watchers is that the market has matured past the 'exciting demo' phase. Successful pilots that never become full deployments are a common story. The prevailing view is that deals now die over operational concerns, not model performance. Enterprise buyers are asking harder questions about implementation risk, governance complexity, and workflow disruption before they commit. Startups still optimising for initial excitement are seen as missing the point. The shift is towards solutions that reduce uncertainty and integrate cleanly into existing systems — a less glamorous but more durable path to revenue.
Sailfish's take
This isn't surprising — it's the natural next step for any new technology inside a large organisation. We've seen this pattern before. The conversation has to move from 'what it can do' to 'how we can run it'. For builders, this means your pitch deck needs a major overhaul. We'd argue the technical appendix is now more important than the hero demo. You need slides on governance, integration paths, and what happens when the model fails. If you're selling to the enterprise, you're no longer selling a model. You're selling operational stability.