As finance leaders chase shorter closes and smarter insights, Stacks’ founder Albert Malikov believes AI can finally flip the month-end equation, automating 90% of the grind while keeping accountants in control.
In Partnership With
For most finance leaders, the month end close is still the least automated, most time-consuming process in the business. It’s where teams spend days cleaning data, matching transactions, and fixing errors that rule-based tools continue to miss. For many CFOs, it’s the one process that never seems to scale.
Stacks, founded by former Uber product lead Albert Malikov, is trying to change that. The company’s pitch is simple: use AI to take on the repetitive, rules-based work that consumes 80 percent of the close, while keeping finance professionals firmly in control. We asked him where the vision came from.
“Ten years ago, at Uber we were scaling across 60 regions, billions of transactions, dozens of business lines” … “We tried to automate as much as possible, but the logic and complexity made it hard. We needed custom code for everything. It worked, but it took a huge engineering effort”. That experience shaped his belief that the missing piece wasn’t process discipline, it was technology. With modern AI, he argues, companies no longer need armies of engineers to achieve a fast, accurate close.
The modern month end can be divided into three main phases.
First comes data and ETL—transforming raw transactions into categorized, usable data. Revenue and cost data must be tagged with the right accounts, departments, and metadata. Even in mature organizations, this remains largely manual. Stacks estimates that roughly 40 percent of month end time is spent here.
The second phase is reconciliation—the painstaking work of checking balance sheet accounts, from cash and bank reconciliations through to AR and AP. This typically consumes another 40 percent of the cycle.
Only the remaining 20 percent of time goes to reporting and insight—the work that actually informs business decisions. In an ideal world, that ratio would be reversed: 80 percent of time spent understanding performance, 20 percent maintaining the data that underpins it.
Accounting, says Malikov, is built on exceptions rather than rules. He offered a simple example:
“Let’s say your company buys from Amazon. Normally, you purchase cloud hosting services, so your finance teams’ rule books those transactions to Cost of Goods Sold (COGS). Then your marketing team purchases event supplies from Amazon. Same supplier, but a completely different expense category. That one rule no longer works. Someone must go in and change it. Multiply that by hundreds of thousands of transactions and you see the problem.”
For Malikov, this is why older rule-based systems often fail. “They can’t understand context. The world changes faster than the rules.”
Stacks’ approach is built around the principle of human-in-the-loop control. Finance functions have zero tolerance for error, so the system handles the heavy lifting—preparing, categorizing, and reconciling data, while accountants review and approve the results.
“The human is always making the final review,” he said. “We are not replacing accountants. We’re giving them superpowers to work faster.”
Rather than deciding which tasks are human and which are automated, Stacks’ model automates everything that involves manual preparation and presents the results transparently for review. “We automate over 90 percent of the work,” said Malikov, “but still meet the highest standards of accuracy and accountability. The system does the work, and humans approve it.”
Traditional automation takes months to implement because every rule must be written manually. Stacks take a different route.
“When we connect to your ERP, we instantly have access to years of historical decisions,” Malikov explained. “AI can recognize the patterns in that data — even those made by team members who’ve since left. So, from day one, the system already understands how things were done.”
It also learns continuously. “When a user corrects a suggestion — for example, changes a categorization, the system remembers that and applies it next time. It’s getting smarter every day by observing your team’s work.” That approach shortens what Malikov calls “time to value.” “Legacy systems might take nine months to go live. For us, it’s days for smaller companies and weeks for complex ones.”
So where should finance leaders draw the line between AI and human oversight?
“All data manipulation: preparing, categorizing, cleaning — should be handled by technology,” said Malikov. “Machines do it faster and with fewer mistakes. But reviewing, approving, and providing business context must stay with humans. They understand the company, compliance, and risk.” That, he argues, is how teams move from spending their days in spreadsheets to focusing on insights that actually steer the business — decisions like which products to continue investing in, or where to reallocate resources.
Before posting anything automatically, Malikov says CFOs should insist on three safeguards:
If Stacks had 30 to 60 days with a finance team, Malikov said he would measure two things:
“First is how quickly the system can be implemented — how fast we can get from connection to automation. Second is what the team does with the time they get back. If they start bringing insights to leadership that they didn’t have time to find before, that’s success.” He shared one example: “If a controller comes back to leadership and says, ‘I finally had time to dig into a business line and found its loss-making,’ that’s the right outcome,” says Malikov. The point is not just to close faster, it’s to think faster.
Technology only works when behavior changes with it. The shift starts with leadership direction: CFOs setting the tone that finance will be modern, efficient, and insight driven. It also requires the right tools: not generic chatbots or productivity apps, but secure, purpose-built finance workflows that align with regulatory and audit requirements.
Finally, expectations must evolve. “Close faster” is no longer enough. The goal is richer insight, better foresight, and a more strategic finance function.
The idea of a “one-click close” has become a buzzword in finance tech, but for Malikov it’s a practical goal.
“To get there, you need a continuous close,” he said. “Transactions come in real time, and ETL and reconciliations happen throughout the month. When that’s true, the close becomes reporting.”
Stacks’ latest Agentic AI platform already handles transactions continuously and automates most balance sheet reconciliations, with wider rollout underway. For simpler, digital-first businesses, the company believes the target is already within reach. More complex operations, such as those dealing in physical goods, will take longer—but the horizon is measured in quarters, not years.
The biggest impact, Malikov says, comes in companies with established finance teams — typically ten to fifty people working on the close, often within larger businesses of 300 to 10k employees.
“When you drive 60 to 70 percent productivity gains in a team of 60 people, that’s a massive impact,” he noted. Most Stacks clients already use ERPs like NetSuite, Microsoft Dynamics, or SAP.
At that scale, the payoff from automation is clear. “If teams can reclaim that time and focus on strategy instead of spreadsheets,” said Malikov, “you’re not just closing faster, you’re building a smarter finance function.”