The financial close, reimagined - by people who actually had to close the books.
An AI-native platform for the office of the CFO. It does the most thankless work in finance - reconciliations, journal entries, the month-end close - so the humans can go home.
Exhibit A: the Stacks calling card. The kind of logo that wants to make spreadsheets feel obsolete - and very nearly does.
Somewhere in a finance department right now, it is 11pm on the final day of the month and the lights are off. That is the strange part. The close is happening anyway.
For most of corporate history, the month-end close has been a ritual of caffeine and dread: a finance team stitching together spreadsheets, chasing down reconciliations, and explaining variances long after everyone else has logged off. Stacks decided that ritual was optional. Its software runs the close as an agentic workflow - reconciling transactions, drafting journal entries, and writing variance commentary that can actually explain itself.
Today Stacks is a London-based company of around 44 people, backed by a $23M Series A led by Lightspeed, with more than 30 enterprise finance teams trusting it with the part of their job they like the least. It is not a dashboard. It is the thing doing the work.
Sales got a CRM. Engineering got the cloud. Marketing got more dashboards than it knew what to do with. Accounting got Excel and a deadline. The month-end close - the process of squaring every number a business produced - still runs largely on manual matching, copy-paste, and the institutional memory of whoever has been there longest.
The tension is simple. Finance is supposed to be the function that tells a company the truth about itself. Instead, it spends most of its days assembling that truth by hand, leaving little time to interpret it. The close eats the analysis.
Stacks exists to resolve that tension. If the mechanical work - the reconciliations, the postings, the variance write-ups - can be handled by AI agents that keep a clean audit trail, then the people can do the part only people can do: decide what the numbers mean.
Albert Malikov was a product lead at Uber. The founding team is drawn from Uber, Plaid, Mollie and Miro - companies that spent the last decade making consumer and developer products feel effortless. Their bet was that the same principles could be pointed at the least glamorous corner of finance, and that the timing finally made sense.
The wager rests on a less obvious idea than "add AI to accounting." Stacks argues that AI agents can only be trusted in finance if they sit on an AI-ready data layer - a clean, structured foundation under the ledger. Build that layer first, and the agents become reliable instead of merely impressive. Skip it, and you get a very confident robot making journal entries you cannot audit.
It is a contrarian place to plant a flag. Most AI startups chase the visible, demo-friendly tasks. Stacks chose bank reconciliations - the work nobody brags about - precisely because that is where the hours, and the trust, are won.
Albert Malikov and a team from Uber, Plaid, Mollie and Miro start building an AI-native finance platform.
Consecutive rounds led by EQT Ventures and General Catalyst to simplify the financial close toward a single click.
Automated variance analysis replaces spreadsheet commentary with explainable, account-level investigation.
The CFO profiles Stacks as AI's one-click close finally moving from pitch deck to practice.
Led by Lightspeed, with EQT Ventures, General Catalyst and S16VC participating. 30+ enterprise customers onboarded.
Stacks is not one feature; it is the close, broken into agents. Each piece takes a slice of the work a finance team used to do by hand.
Tracks tasks, dependencies and notifications across the whole close cycle, with an audit-ready trail behind every step.
An AI-native engine matches high volumes of transactions and flags discrepancies - automating up to ~95% of reconciliations in some deployments.
Prepares and posts journal entries automatically. Nivoda reports postings dropping from days to minutes.
Automates variance analysis with explainable, account-level investigation instead of hand-typed spreadsheet notes.
Generates executive summaries with real-time trend and variance insight for the people who read the close, not run it.
Plugs into systems like NetSuite and syncs with Excel, so the data layer fits the stack a company already runs.
Skepticism is the correct default for any company promising to automate accounting. So here is the evidence, in the only language a finance team respects.
// reported customer outcomes · lower is the point
Bars sized for readability, not to the third decimal - finance teams can do their own variance analysis. Sources: Stacks customer reports (Juni, Nivoda) and company figures.
Spend-management scale-up using Stacks for its finance operations.
Switched from a legacy provider and saw automation benefits within weeks.
Journal postings went from days to minutes; ~95% of reconciliations automated.
Among the 30+ enterprise teams running their close on Stacks.
Lightspeed led the Series A, with EQT Ventures, General Catalyst and S16VC doubling down from the earlier rounds. Investors who back the same company twice are, in their own understated way, paying a compliment.
The stated goal is bigger than faster reconciliations. Stacks is aiming at the roughly $100B office-of-the-CFO software market and, beyond it, the far larger labor spend still powering enterprise finance by hand. The thesis: if the data layer is clean and the agents are trustworthy, the close becomes infrastructure rather than an event.
What that buys back is attention. A finance team that is not manually closing the books is a finance team that can actually analyze them - the difference between a function that records the past and one that helps shape the next quarter.
There is a lot of speculation about which white-collar work AI changes first. The flashy guesses get the headlines. The likelier answer is duller and closer: the high-volume, rules-heavy, audit-bound work of operational finance. It is exactly the shape of problem agents are good at, and exactly the work humans are happiest to hand over.
If Stacks is right, the month-end close stops being a milestone on the calendar and becomes a background process - always running, rarely thought about. The skeptic's fair question is whether the audit trail and accuracy hold up at enterprise scale across years, not quarters. That is the bet the next chapter has to prove.
Back to that dark finance office at 11pm. The lights are off not because the work was skipped, but because it was already done. The close happened. Nobody had to stay. That is the whole idea - and increasingly, it is just Tuesday.