Source: Forbes
Zalos, a computer-agent startup focused on finance operations, is announcing a seed round this week, adding another data point to what has become one of the most active investment themes in enterprise software: agentic AI for the CFO tech stack.
The San Francisco-founded, Y Combinator-backed company plugs directly into the financial systems clients already use. This includes ERPs, accounts payable platforms, and CRMs, upon which Zalos then autonomously executes multi-step workflows that otherwise consume thousands of dollars in labor hours each month. For a mid-size company with four accounting staff, manual reconciliation alone can run $67,200 a year in direct labor costs, according to Ramp. Billing operations, month-end close tasks, and reconciliation run around the clock, without the overhead of human headcount.
“CFOs get fired for two main reasons,” commented Will Fairbairn, co-founder of Zalos. “They commit fraud, or they have a bad ERP migration. Almost every CFO I met complained about their ERP. That made me realize how painful it will ever be to rip these systems out.”
Rather than asking finance teams to replace their infrastructure, Zalos works within it, using what Fairbairn describes as a “defense in depth” approach: breaking workflows into discrete tasks, running vision-based evaluations after every action, and deploying self-healing agents that can troubleshoot in real time when something goes wrong.
Why Finance Teams Are Finally Ready to Adopt AI
For years, skepticism around AI ran high. Now, with reliability improving and audit-ready outputs becoming table stakes, the appetite to experiment has changed meaningfully.
Xero, the cloud accounting platform serving small businesses and their advisors, is making a similar argument from a different vantage point. Xero’s Chief Product and Technology Officer Diya Jolly framed the company’s AI direction around what she calls “Accountable Intelligence,” describing a model where automated systems show their work at every step while humans retain final decision authority. “True intelligence should be accountable, not artificial,” Jolly wrote, articulating a standard that larger competitors and newcomers alike are now racing to meet.
Who Owns the Finance Context Layer?
For Zalos, the roadmap extends well beyond automating individual tasks. Fairbairn’s longer-term vision centers on building what he describes as a finance context graph across all of a company’s systems, positioning Zalos as the connective layer through which any future finance agent operates. The company is also differentiating on pricing, charging per agent rather than per seat, a direct challenge to the incumbent ERP model that will force larger players to adapt.
“The ceiling of what is possible is just mind blowing,” said Zalos CTO, Hung Hoang. “Finance and accounting is rule-based. They are verifiable workflows. These are some of the most important ingredients for agents, enabling techniques like reinforcement learning.”
Whether the market consolidates around a few platforms or fragments into dozens of point solutions remains an open question. Inside the CFO’s office, the transformation has already begun.
