TL;DR: Capital is pouring into AI-native tools for each piece of the office of the CFO: spend, the general ledger, and revenue. Abacum CEO Julio Martínez argues that faster pieces don't add up to better decisions on their own, and that designing how the pieces connect to the plan is now the CFO's job.

Three deals, one pattern

Last week I sent a note to my whole team about three pieces of news. Ramp raised $1.85 billion at a $60 billion valuation. Intuit agreed to acquire Sequence. And Rillet became a unicorn, raising at a $1 billion valuation.

Ramp, Rillet and Sequence are partners and friends, so I was genuinely happy for them. But what interests me more is what the three deals have in common. Each one is a bet on a single piece of the office of the CFO. Ramp started with cards and has moved into payments, procurement, vendor management and, lately, accounting. Rillet is rebuilding the general ledger and the close. Sequence automates quote to revenue, which its CEO calls "the most fragmented workflow in the back office."

Serious capital is going into making each piece AI-native, and it's working: spend gets approved, books close, and invoices get collected faster. If you run finance at a company of a few thousand people, your stack two years from now will probably be a set of very good, very fast specialist systems.

The obvious reading is that finance is getting automated, one piece at a time. I think that reading is incomplete.

Why a faster finance stack doesn't guarantee better decisions

Each of these tools is excellent at its own piece. None of them is necessarily responsible for whether the pieces add up. Say marketing commits $400K to a new agency, and in the same month a large customer moves from seat pricing to usage pricing. The spend tool knows about the agency. The billing tool knows about the pricing change. The ledger will know about both at month-end. No single system knows whether that combination still fits the hiring plan the board approved in January.

That question lives between the systems, and today a person usually answers it. Someone in FP&A exports from three places, reconciles in a spreadsheet, and walks into the forecast review with an answer that's two weeks old. I've called this the Human API before: finance as the layer every question has to route through.

Here's the part I think gets missed. When every piece of the stack was slow, a slow person in the middle was fine, because everyone was waiting for month-end anyway. When spend, revenue, and the close move at machine speed, the middle person becomes the slowest part of the system. By the time finance sees the problem as a variance, the spend tool and the billing tool both knew weeks ago.

How to map your finance stack in one sitting

Here's an exercise for budget season. Draw your office of the CFO as boxes: spend, the ledger, revenue and billing, payroll and HR, the plan. Then for each pair of boxes, ask two questions.

Which decision depends on both? Spend and the plan meet at hiring and vendor commitments. Revenue and the plan meet at pricing changes and the cash forecast. Payroll and the ledger meet at every headcount variance you've ever had to explain to the board.

What connects them today, a system or a person? If it's a person, write down who, and how many hours a month it takes them. That list is your real integration roadmap, and it's usually shorter and more specific than the one sitting in a project tracker.

The CFO's job is designing how the pieces connect

The CFOs I talk to who are getting this right don't describe themselves as running finance workflows anymore. Specialist software runs the workflows, and runs them well. Their job has become designing how the company's financial picture holds together: which numbers are shared, who owns the definitions, how fast a change in one box shows up in the others. Finance becomes the architect of how the company thinks.

So here's the question I'd put to your team this week. If every piece of your stack got twice as fast tomorrow, would your company make better decisions, or just faster ones?

How Abacum helps

Abacum is the AI-native FP&A platform where the pieces of the finance stack meet the plan. It connects to the systems finance teams already run, including Ramp for spend, Rillet for actuals, and NetSuite, Stripe, Salesforce and HRIS tools, so the forecast reflects what the spend tool, the ledger and the billing system already know.

Highway, a freight-technology company, built its finance stack this way: Rillet as the accounting foundation, Ramp for spend, Stripe for billing, Salesforce for the commercial record, and Abacum for planning. Read how Highway connected five finance systems and got month-end reporting down to days, not weeks.

And when a number moves, Abacum Intelligence shows exactly which data sources, formulas, and assumptions drove any AI-generated result, so the team can trace a change back to the system it came from.