Vertical AI for banking just hit $26M: Gradient Labs bets autonomous agents beat co-pilots on compliance

Gradient Labs has expanded its Series A funding round to $26 million, adding Octopus Ventures and CommerzVentures as new lead investors, with follow-on participation from Redpoint Ventures and Exceptional Capital. The London-based startup, which builds AI agents to automate regulated workflows in financial services, now reaches more than 32 million end users and recorded 900% revenue growth over the past twelve months.

What does the $26M mean for banking operations?

The capital doubles the company's original $13 million Series A, closed in July 2025, and will fund a transatlantic expansion into the US market alongside continued product development. The round reflects a widening investor conviction that finance-specific, or "vertical," AI can solve a problem horizontal platforms have struggled with: automating complex, multi-step regulated workflows without creating compliance exposure.

Gradient Labs was founded in 2023 by Dimitri Masin, Neal Lathia, and Danai Antoniou, all former employees of Monzo, where they led AI and data science work across customer operations and financial crime systems. The founding thesis was that general-purpose AI models, however capable, cannot meet the domain-specific compliance, audit, and edge-case demands of regulated financial institutions. Three years in, the business argues the numbers support that view.

How does the agent suite actually work?

The company has moved beyond a single-agent architecture to a suite of specialist agents, each purpose-built for one regulated workflow. In production today, the Lending Agent automates the full borrower lifecycle, from a missed payment through outbound collections to an agreed repayment plan. A Disputes Agent handles chargeback intake, investigation, and resolution. A KYB Agent runs identity and document verification. The agents operate as an interconnected system, sharing context and memory and handing off work between each other when a customer journey crosses domains.

Current customers include Wise, Monzo, Zego, and Pockit in the UK and Europe, and US neobanks Current, Stash, and Rho. Gradient describes its agents as compliant with both FCA Consumer Duty and the EU AI Act, with domain-specific guardrails and test scenarios built into each agent at the architecture level.

Voice AI in financial services, widely regarded as the most difficult channel to automate safely, is in live production at scale. The company claims it is one of the only vendors running this capability in a regulated environment, and frames voice as a key competitive moat against broader enterprise AI platforms such as San Francisco-based Sierra, which raised $950 million in 2025, and Decagon, which raised $250 million, both of which target large enterprises across sectors rather than focusing on finance specifically.

What do the performance figures say about autonomous versus co-pilot AI?

The debate between full autonomy and AI-assisted human agents (the "co-pilot" model) has been a defining tension in enterprise AI deployments. Gradient takes the minority position: that full autonomy, not co-piloting, produces better compliance and customer experience outcomes. In every customer deployment, the company reports customer satisfaction scores above human baselines, with some reaching 98%, and resolution rates of 80 to 90 percent across long-running processes. The company backs that claim with an unusual commercial guarantee: if a deployment fails to perform as scoped, customers receive a refund.

This performance-first positioning has clear investor appeal. Mordor Intelligence estimates the agentic AI market in financial services at $5.51 billion in 2025, growing to $33.26 billion by 2030 at a 43% compound annual rate. A McKinsey analysis cited by market researchers puts a 30% probability on AI substantially reshaping global banking workflows within this decade, with up to $170 billion in annual banking profit at risk for institutions that do not adapt. Against that backdrop, the transatlantic composition of Gradient's cap table, led by investors who have backed category-defining fintechs on both sides of the Atlantic, signals that vertical AI infrastructure for financial services is being treated as a serious asset class, not a niche product category.

What comes after the Series A?

Gradient's stated roadmap is to build what it describes as an "operating system" for autonomous banking operations: a layer that ties specialist agents together, withstands regulatory scrutiny, and earns institutional trust at the level where a bank's licence is on the line. The US expansion, now underway with customers including Current, Stash, and Rho, is the first major test of whether a compliance architecture designed primarily around FCA and EU AI Act frameworks can translate cleanly into the US regulatory environment, where financial oversight sits across the OCC, CFPB, and state-level regulators simultaneously.