Inside Payhawk’s Shift to AI-First Finance
Payhawk’s CEO, Hristo Borisov, is clear about the company’s fresh approach to AI in finance. Instead of grand promises, the focus is sharp: helping finance leaders cut routine tasks without losing control. “Finance leaders don’t need another assistant that parrots policy,” he says. “They need a system that understands context and gets things done, under their control.”
The newly launched Financial Controller Agent is at the heart of this strategy. It’s the first in Payhawk’s AI Office of the CFO suite, with three more agents coming in September. Unlike typical chatbots that just summarize or search documents, this agent takes action. It pursues missing invoices, flags incomplete transactions, suggests fixes, and enforces company policies—all within the finance team’s set boundaries. The goal isn’t to replace CFOs but to free them from manual work that distracts from strategic priorities.
From Admin to Execution
Most finance teams still spend too much time on routine tasks that don’t align with what’s expected of modern CFOs. Borisov points out that CFOs should focus on capital allocation and spotting market opportunities, yet legacy tools pull them into tedious operations. He sees AI’s role as bridging this gap—but only if it’s built into the workflow, not added as an afterthought.
Payhawk’s agent is designed around workflows, not general knowledge or corporate speak. It’s procedural, rules-based, and interacts directly with spend management processes. “This is not another experiment,” Borisov says. “It’s a working layer of intelligence that lives inside the core workflow.”
What Counts as an Agent?
Borisov critiques how the term “AI agent” has been diluted. Many so-called agents merely display information in a chat window. “There’s a difference between something that can respond to a prompt and something that knows what to do next,” he explains.
The Financial Controller Agent operates continuously in the background. It learns from user actions, highlights risks, suggests changes in categorization, and escalates policy breaches. Importantly, it acts without a prompt but keeps the finance team in control. It never makes silent changes; every action is documented, auditable, and reversible. “There is no AI without auditability,” Borisov stresses. “We’ve built this for finance, not for general productivity.”
A Strategic Pivot
For Payhawk, launching the agent isn’t just a new feature—it signals a new direction. “We are now an AI-first company,” Borisov says. This isn’t marketing speak but a clear technical path. Instead of retrofitting AI onto old systems, Payhawk is focusing on domain-specific solutions built with the rules and realities of finance in mind.
The market is changing. Established ERP providers are layering AI on top, while startups compete on user interfaces. Borisov believes there’s a middle ground: a product that understands finance logic and is wired to execute, not just show data. “Speed, compliance, visibility – that’s the triangle finance teams live in,” he says. “If we can help maintain all three without them lifting a finger, that’s real value.”
Built for CFO Constraints
Borisov avoids hype. He doesn’t promise AI will transform corporate finance overnight. But he’s confident the right, quietly effective approach can cut much of the inefficiency that clogs day-to-day work. The Financial Controller Agent isn’t a chatbot or virtual assistant. It’s a background system handling repetitive spend management tasks while keeping CFOs fully in charge.
As more vendors rush into AI, Payhawk’s approach stands out for its restraint and focus. It doesn’t seek headlines with bold claims. Instead, it offers a practical example of how automation can support decision-makers without replacing them. In a market crowded with demos, this system does less talking and more doing.
For finance professionals interested in practical AI tools and automation in their workflows, exploring domain-specific solutions like Payhawk’s agent may offer a clearer path forward.
Your membership also unlocks:
