Why the cash conversion cycle keeps drifting, why dashboards can’t fix it, and why the platforms racing to buy AI for the enterprise still haven’t touched finance operations.
Ask your finance organization four questions. Can you see where every dollar is stuck right now? Not at close. Not on a lag. Can you act on that before it hits cash? Do you trust your 30/60/90-day forecast? And when the board asks who owns the result, is there one name, or four names each pointing at a different leg of the same cycle?
Most enterprises can’t answer yes to all four. Our CEO, Rohit Gupta, has spent the past several weeks on his Substack making the case for why. Read together, his last few posts add up to something sharper than anyone argues alone. The ownership gap in the cash conversion cycle isn’t a discipline problem. It’s a missing architecture, and the architecture that closes it is already buildable.
The cycle nobody owns.
In “The Cash Conversion Cycle Nobody Owns,” Rohit’s framing starts simple: the cycle runs every business, but nobody runs it. AR owns collections, AP owns payments, Treasury owns cash, and Procurement owns spend. Each has a clean view of its own leg, while the loop itself still advances one email, one dispute, one approval at a time. A discount lapses because an invoice sat in approval for 3 days longer than necessary. A customer quietly stretches from 30 to 45 days. A dispute festers for weeks. None of it shows up in a board deck; all of it shows up in working capital.
His four-question diagnostic is visibility, control, predictability, and accountability. Data back it most CFOs would recognize: PwC finds more than a quarter of global cash invisible to Treasury on any given day, Ardent Partners puts straight-through invoice processing at barely a third industry-wide, and Strategic Treasurer found 53% of Treasury teams now call forecasting “difficult,” up from 39% a few years ago.
That’s a precise diagnosis. But it’s only a diagnosis. What actually closes the gap is where the rest of Rohit’s writing this quarter picks up.
Why a dashboard was never going to fix this.
In “SmartResearch: The Intelligence Layer the Cash Cycle Has Been Missing,” Rohit reframes the problem: finance doesn’t have a data shortage, it has a synthesis shortage. The ERP tells you what happened; it was never built to tell you what to do about it. And no single AI model closes that gap alone. Rules are deterministic and auditable but rigid. Machine learning surfaces patterns but can’t explain itself to an auditor. LLMs synthesize plain-English answers but, left alone, produce fluent, confident, wrong ones.
The architecture that works coordinates all three: rules, ML, and LLM synthesis. Every answer carries source lineage back to the invoice or ERP line that generated it. Map this onto the four questions: visibility becomes a live cross-leg read of AP and AR, control becomes same-day exception surfacing, predictability becomes a forecast that’s a byproduct of real payment behavior, and accountability finally has a single auditable view to point to. SmartResearch isn’t a different answer. It’s the mechanism that the ownership piece never names.
Why has nobody built this yet?
So why hasn’t a major platform shipped this already? In “The Platform Wars Are Coming for Finance,” Rohit points out that over the past 15 months, enterprise software has spent roughly $4 billion acquiring AI specialists: Workday/Sana, ServiceNow/Moveworks, SAP/Dremio. Every deal targets HR, IT, or data infrastructure. Finance operations, despite a nearly $11 billion market, are conspicuously untouched.
The reason: finance workflow intelligence can’t be acquired the way HR intelligence can. An exception in AP is an overpayment or compliance failure, not a missed alert. The margin for silent failure is near zero, and the intelligence to handle that safely comes only from millions of real transactions processed in production, not a foundation model. That’s a moat built one deployment at a time, not bought in an acquisition. Whoever builds it now owns the category when consolidation eventually arrives.
The four questions are a vendor checklist.
Here’s the reframe: none of the three posts states outright that the four questions from “The Cash Conversion Cycle Nobody Owns” aren’t just a self-diagnostic. They’re a checklist for any vendor selling “AI for finance.” Does it show you a live picture or a stale report? Same-day exceptions or a monthly review? A forecast grounded in real behavior, or a spreadsheet with an AI label? And can every number be traced back to a source without calling IT?
Most platforms answer two of the four convincingly. The cash cycle will keep running in every business, whether or not anyone owns it.
The only question is who gets there first.