Artificial intelligence has evolved in clear phases. Generative AI defined 2023, multi-modal systems expanded capabilities in 2024, and reasoning models pushed boundaries in 2025.
Now, Computer Agents are emerging as the defining shift for 2026 — software that can operate systems rather than simply assist users. While companies like OpenAI and Anthropic are building general-purpose agents, San Francisco-based Zalos, founded in 2025, is focused entirely on finance, where precision, traceability, and auditability are critical.
Today, the startup has raised $3.6 million in seed funding to bring Computer Agents into finance. The round was led by 14 Peaks, with participation from Cohen Circle and 20VC.
Angel investors include Mike Lenz (CFO Fedex), Ian Sutherland (CFO Tide), Long Dinh (CFO Ada), Nancy Casey (Global Vice President, Oracle, SAP), Paul Forster (Founder, Indeed), Henri Stern (Founder, Privacy), Ed Woodford (Founder, zerohash), James Beshara (Founder, Tilt Payments), Long Lu (Founder, Misa Accounting), Catherine Dahl (Founder, Beanworks Accounts Payable), Pablo Palafox (Founder, Happy Robot), Hasan Sukkar (Founder, 11x), Chris Smoak (Founder, Atrium), Ooshma Garg (Gobble), Minh Pham (Head of Browser Infra, Perplexity), Jon Langbert (Founder, Alight), Mandeep Singh (Founder, Trouva), Thai Duong (Founder, Calif), Ash Rush (Founder, Sterling Road), Jake Klamka (Founder Insight Data Science), Jonathan Meeks (Board, TA Associates).
Turning workflows into autonomous agents
Zalos automates finance work by learning directly from screen recordings. A workflow performed inside an ERP or Excel can be converted into an agent that replicates the same process.
These agents log in to systems, navigate interfaces, enter data, and validate outputs across ERPs, spreadsheets, email, and internal tools. They already operate on platforms such as NetSuite, Sage, and SAP S/4HANA.
Every action is captured in an auditable log, supported by enterprise-grade security including single sign-on, role-based access controls, SOC 2 Type II compliance, and on-premise deployment options.
This enables automation of tasks such as billing, reconciliations, and reporting, processes that typically require manual coordination across systems.
Solving the “human API” problem
Finance teams operate across disconnected systems that rarely integrate smoothly. As a result, professionals spend significant time manually moving data, reconciling discrepancies, and stitching together reports.
Zalos was built on the idea that the next leap in productivity won’t come from replacing these systems, but from software that can operate them as humans do.
Detailing the origin of this idea, William Fairbairn, CEO & Co-Founder of Zalos, told TFN their background and the idea behind this startup.
While at Agicap (CFO-focused software), he spoke with hundreds of CFOs; the most consistent feedback was their deep hatred of their ERP implementations – these take 12+ months, with little upside if they go well, and significant career risk if the implementation fails. Zalos was built on the insight that AI automation must sit atop the existing CFO stack, not replace it.
In parallel, CTO Hung Hoang had left Apple Pay after 5 years and became fascinated with Computer Agents, including working at a top lab at the time focused on Computer Agents (Twin), which seemed like the true form of AGI. Hung loved Computer Agents because they avoided relying on APIs, which are often unavailable or very problematic to work with in Finance Systems. So they began work on Zalos, Computer Agents for Finance Operations, last October when they joined Y Combinator (Fall 2025 batch)
Their specialised Computer Agents operate Finance Systems like a human (via login), avoiding heavy integrations and enabling radical levels of human emulation of Finance teams.
Built to work the way finance teams do
At the centre of Zalos is a product designed to mirror how finance actually works. Instead of asking teams to rebuild workflows or integrate new systems, it starts with what already exists.
The platform converts screen recordings of finance processes, whether inside ERPs or Excel, into fully functional Computer Agents. A recorded billing cycle, reconciliation process, or reporting task becomes a repeatable, automated workflow.
These agents operate across the entire stack: ERPs, spreadsheets, email, and internal tools. Logging in with credentials, navigating interfaces, clicking through dashboards, entering data, and checking outputs against predefined controls, just as a finance professional would.
Every action taken by an agent is recorded in a detailed, auditable log. This ensures full traceability, something finance teams cannot compromise on.
Alongside this, the platform is built with enterprise-grade security at its core, including single sign-on, role-based access controls, SOC 2 Type II compliance, and options for on-premise deployment.
Designed for CFOs
Zalos is built with a clear user in mind. At the top are CFOs and finance leaders, the decision-makers responsible for accuracy, efficiency, and risk.
“Finance teams have the systems, but they are still doing the work manually because the stack is not connected. We built Zalos on the belief that CFOs should not need to rip out their existing stack to adopt the latest in AI. We want to start by building on what is already there. Computer Agents that can log in and run the workflow end-to-end are the fastest path to real transformation in finance operations,” said Fairbairn.
On the ground, it serves mid-market and enterprise finance teams managing complex, multi-system operations. These are teams already juggling ERPs, spreadsheets, and fragmented tools daily.
Beyond traditional enterprises, the platform is also gaining attention from tech companies and private equity firms. For PE firms in particular, the appeal is direct: improving operational efficiency across portfolio companies without forcing costly system overhauls.
In environments where multiple systems are already in place, Zalos offers a way to connect and operate them without having to start from scratch.
Competition and what comes next
Zalos operates in an emerging category alongside generalist Computer Agents. Competitors include OpenClaw, now acquired by OpenAI, and enterprise-focused tools like Anthropic’s Claude CoWork.
Looking ahead, Zalos plans to expand deeper into enterprise ERP environments and on-premise systems. Its broader vision is to build a connected layer across the finance stack, enabling multiple agents to work together seamlessly.
If successful, finance teams could deploy networks of agents handling workflows end-to-end, reducing manual effort while improving speed and accuracy.
Regarding future plans, Fairbairn said, “Zalos already has customers across the major midmarket ERPs (including NetSuite, Sage, SAP S/4HANA). We’re excited to expand to enterprise ERPs and more on-premises systems. Building this wide-reaching context graph for the Finance team will enable us to be positioned to support our CFO customers with radical levels of human emulation by Zalos agents.”
