Gradient Labs Raises $13M to Bring Safe AI Automation to Regulated Industries

Gradient Labs, a London-based AI startup building deeply specialized customer service agents for regulated industries, has raised $13 million in Series A funding. The round was led by Redpoint Ventures, with participation from Localglobe, Puzzle Ventures, Liquid 2 Ventures, and Exceptional Capital. The funding signals a growing demand for AI systems that go far beyond superficial automation—and instead embed regulatory intelligence, procedural logic, and auditability directly into customer operations.

The capital will accelerate Gradient’s product development and hiring across engineering, research, onboarding, and go-to-market teams. More significantly, it will fuel the company’s broader mission: solving the operational strain facing regulated industries through a new class of domain-specific AI agents.

The AI Challenge in Regulated Sectors

Customer service in finance, insurance, and other high-risk industries is uniquely difficult. On the one hand, customer expectations have skyrocketed—66% of people now expect a near-instant response, and nearly one in three will abandon a company after a single bad experience. On the other hand, firms in regulated spaces cannot simply plug in consumer-grade chatbots. The risks—from compliance violations to data mishandling—are too great.

Traditional AI tools offer only partial solutions. Most are trained for general-purpose queries, and even the most advanced customer support agents today typically handle only the simplest 20–25% of inquiries. These tools struggle with layered workflows, verification steps, legal nuance, and escalating decision trees. In financial services, this is where the bulk of costs and risks lie.

Gradient Labs addresses this gap directly.

A Founding Team With Domain Credibility

Gradient was founded in 2023 by Dimitri Masin (CEO), Danai Antoniou (Chief Scientist), and Neal Lathia (CTO)—all of whom played critical roles in building the infrastructure and operations at UK neobank Monzo. Their experience gives them an unusually deep understanding of the real-world constraints that regulated firms face: how fraud detection systems are designed, how compliance departments operate, and what internal tooling actually looks like in a high-risk environment.

This founder-market fit is rare, and it shows in the traction Gradient has seen since launch. Within three months, the company secured nine customers—including one of Europe’s largest banks. Clients now report resolution rates of up to 90% and CSAT scores exceeding 98%, numbers that are virtually unheard of in regulated support environments.

What Gradient Labs Actually Builds

At the heart of Gradient’s offering is Otto, a procedural AI agent trained not only on language, but on logic, workflows, and regulation-specific processes. Otto is designed to do more than deflect tickets—it executes complex, multi-step operations with contextual awareness and institutional memory. This includes:

• Authenticating customers based on regulatory KYC logic
• Freezing and replacing lost or compromised cards
• Initiating fraud investigations with audit-trail documentation
• Updating sensitive financial records based on customer intent
• Navigating policies with precision across jurisdictions and use cases

Unlike large language models used in general-purpose tools, Otto is fine-tuned to function as an agent within a system, not just as an interface. It reads and writes data into existing tools like Intercom, Zendesk, and Freshdesk, and operates within strict guardrails. Every action Otto takes is explainable, logged, and reproducible—key requirements for firms under financial regulation.

Deep Automation Without Sacrificing Control

One of the most significant technical differentiators is Gradient’s use of procedural abstraction rather than purely generative reasoning. Where many chatbots guess intent and hallucinate solutions, Gradient’s architecture composes responses and actions from predefined, verifiable steps—similar to a decision engine layered over an LLM core.

This means companies can map out their internal logic (for example, how to handle disputes on a credit card transaction) and let Otto execute it precisely, without human-in-the-loop intervention—but still with oversight. Compliance teams can audit decisions, test edge cases, and impose restrictions, ensuring that the AI remains within approved operational bounds.

And because Gradient’s onboarding process doesn’t rely solely on static datasets, but includes dynamic process learning, resolution rates start high—often 40–60% from day one—and increase quickly as the system adapts to the firm’s exact workflows.

What This Means for the Future of Customer Operations

The implications of Gradient Labs’ work go beyond support tickets. In many ways, the company is building a new AI layer for enterprise process execution, one that is rooted in regulation-aware architecture. Rather than applying AI retroactively to isolated support functions, Gradient is embedding intelligence directly into operational fabric.

This is particularly meaningful for industries that have historically lagged in AI adoption—not because of lack of interest, but because of risk. Financial institutions, for example, are eager to modernize but constrained by internal controls, liability fears, and the need for absolute traceability.

Gradient is offering a viable model for what AI looks like in that context. A model that balances:

• Speed and responsiveness with precision and accountability
• User experience gains with regulatory defensibility
• Deep automation with human oversight and clarity

In doing so, Gradient Labs is helping reshape not just how service gets delivered—but how systems get trusted. If Otto and agents like it continue to succeed, we may look back on Gradient Labs as one of the first real examples of AI not just acting smart, but acting responsibly inside some of the most sensitive institutions in the world.

And that may be the breakthrough that finally brings true AI transformation to the heart of the economy.