Fassen Sie diesen Blogbeitrag wie folgt zusammen:
The Best AI in Finance Conversations Are Happening in Rooms You're Not In
And one of them happened in New York on May 20, 2026.
The Best AI in Finance Conversations Are Happening in Rooms You're Not In
And one of them happened in New York on May 20, 2026.
If you weren't there here's what happened.
Editor’s Note: Polestar Analytics and Databricks co-hosted an invite-only executive dinner in New York "Unlocking Working Capital Intelligence with Agentic AI" bringing together fifty-plus finance, data, and AI leaders from over thirty organisations for one evening, in one room, on purpose. Just talking about the actual truth of where and how AI is working.
Cloud had "digital transformation." Big Data had "data-driven." AI has, well, "AI" the buzzword so good it didn't even need a rebrand. These phrases aren't lies, they're just said so often they've become white noise.
Sit through enough conferences and you'll hear the same three sentences on a loop: AI will transform finance, autonomous systems are the future, everything is changing forever.
Sure. But the interesting conversation tends to start the moment the keynote ends and someone in the hallway asks the question: "Cool, but did the autonomous claim actually work?" Nobody puts that on a conference banner, but it's these questions that decide whether AI graduates from pilot purgatory or dies quietly in a slide deck nobody opens again.
Amit Singh, Global Head of Partner GTM at Databricks, opened with what he called the Working Capital Intelligence Gap which, translated from corporate, means: companies don't have a data shortage, they have a data trust shortage. Fragmented, siloed, and about as coherent as a group chat with forty unread messages.
His argument, governance and a solid data foundation aren't boring back-office plumbing anymore, they're the competitive edge. Building the next model is exciting. Fixing the data behind it is not. Guess which one enterprises actually need, and guess which one they'd rather skip.
As Lakebase and Databricks Genie partners for finance, Polestar Analytics and Databricks are built to solve exactly this Databricks providing the data foundation, Polestar Analytics the expertise to turn that foundation into answers finance teams can actually use.
Mohana Bhrugubanda of Polestar Analytics got into the problems finance teams face every day
- Cash flow forecasting: because guessing isn't a strategy, even when the spreadsheet looks confident
- Treasury visibility: knowing where your money actually is, in real time, not three days later
- AR/AP automation: getting paid faster, paying smarter, without a team manually chasing every invoice. Agentic AI can push straight-through processing past 95%, which is the finance-team equivalent of finally getting your inbox to zero
- FP&A: planning that reflects reality, not last quarter's assumptions. Right now, teams spend roughly 70% of their time just prepping data before any actual analysis begins
- Faster close cycles: because nobody got into finance to spend two weeks reconciling the same numbers. AI-assisted reconciliation can pull that down from 10–15 days to hours
If any of those challenges resonate, you're not alone. They all point to the same need: better intelligence for faster, more informed financial decisions. We explore this in greater detail in our blog on Working Capital Intelligence for CPG Companies and GCCs.
The conversations we had in New York were just one part of a much larger conversation at the Databricks Data + AI Summit 2026. For a broader look at the key announcements, emerging AI trends, and enterprise takeaways from the Summit, explore Databricks Data + AI Summit 2026: Highlights & Key Takeaways.
The CapitalPulse demo on the Databricks Data Intelligence Platform skipped the routine and just showed finance teams asking plain-language questions and getting straight answers instead of a scavenger hunt across six systems. Nobody promised a one-click fix that particular fairy tale has been told, and broken, many times before. The quieter and more believable message: good data plus decent AI equals decisions that don't require a prayer and a spreadsheet.
Wondering what CapitalPulse looks like without the stage lighting? Take a quick tour.
- Moderated by Mohana Bhrugubanda, Industry Head – FSI & Tech at Polestar Analytics, four leaders sat down to talk about what AI in finance actually looks like once it leaves the slide and has to survive a Monday morning: Onur Toprak (VP, FP&A, Campari Group), Pranay Sinha (VP, Enterprise Business Solutions, Cencora), Avijit Roy (CIO, Americas, Moët Hennessy), and Jon Harding (Ex-Global CIO, Conair).
Four themes were addressed:
- AI works best as an advisor, not a replacement. Nobody's ready to let a model near the actual "approve wire transfer" button just yet.
- Trust and governance come before scale not as a nice-to-have bolted on afterward.
- Ethics and accountability aren't garnish. They're not optional seasoning on top of the strategy.
- Better tools alone don't fix the age-old feud between finance and data teams. New dashboard, same trust issues. Shared goals fix that not another integration.
A few things outlasted the dessert course:
- AI isn't the finish line business value is.
- Data governance and Trust isn't a feature you add later; it's the foundation. Skip it, and the AI's recommendations just become expensive suggestions nobody follows.
- The best conversations don't end when the slides do. They carry into dinner, into Monday's meeting, and hopefully into whichever room we all end up in next.
If this left you with more questions than answers, keep pulling the thread:
Every era believes it has finally arrived that this model, this breakthrough, this room, is the one that changes everything. History is kinder and more patient than that. It doesn't reward the arrival. It rewards whoever kept building while everyone else was busy declaring victory.
That's the quieter truth this dinner sat with. Not whether AI will transform finance it will, eventually, in ways nobody in that room can fully predict. But transformation was never really the point. The point was trust: in the data, in the process, in each other. Everything else the models, the demos, the keynotes is just the scaffolding around that one, older idea.
We'll keep setting the table. Pull up a chair next time.
