Round Treasury

Finance teams at growing companies still live largely inside spreadsheets and manual processes: approvals by email, hand-done reconciliations, payroll verified line by line. It is not laziness — it is that the tools designed to automate these operations were built for large corporates, with months-long implementations and costs out of reach for scale-ups. Round Treasury, a London startup, has built a platform that fills this gap with a radically different approach: AI agents that independently handle treasury, payments, and payroll. The core product is an Agentic Workflow Builder that allows CFOs and finance teams to design automated workflows without writing code: from payment approval and disbursement to bank reconciliation and cash flow planning. On top of this layer sits an Autonomous Payroll module that manages payroll disbursement processes on an ongoing basis, reducing the risk of manual errors and cutting processing times. The architecture is designed for the age of AI agents: instead of automating rigid step sequences, Round's agents can adapt to unexpected situations, escalate decisions to humans in ambiguous cases, and learn from corrections. This makes the platform more robust than traditional RPA systems, which break down the moment anything in the expected flow changes. The business model is SaaS, targeting scale-ups and rapidly growing companies that have outgrown the phase where financial operations can be managed manually, but are not yet large enough to justify full enterprise ERP solutions. This market segment — companies with 50 to 500 employees and international operations — has historically been underserved by available tools. Round Treasury represents the convergence of two structural trends: automation of finance processes and the maturation of AI agents as real operational tools. The question is whether it can build a defensible moat before the major incumbent B2B fintech players move in the same direction.