Edinburgh fintech Aveni has raised £12m to accelerate its AI assurance platform for financial services, in one of the larger Scottish fintech rounds of 2026 to date. The University of Edinburgh spinout says the round, led by PXN Ventures as a new investor with continued backing from Puma Growth Partners, Lloyds Banking Group, Nationwide and Scottish Enterprise, will fund new tools designed to monitor the conduct risk of AI agents that deal directly with consumers.
Aveni positions itself as an AI specialist for wealth management, financial advice and banking, building products that sit inside existing customer-interaction workflows rather than replacing them. The company’s technology is aimed at firms that are experimenting with agentic AI for customer service and advice but are constrained by how to evidence that automated conversations meet regulatory expectations. For Scottish financial services players already grappling with Consumer Duty, that makes Aveni less a generic AI play and more a piece of compliance infrastructure.
The new cash will accelerate Aveni’s Unified Assurance Platform, which brings together its existing Aveni Assist and Aveni Detect products with two new offerings called Agent Assure and Agent Approve. The aim is a single framework that can assess and oversee both human and AI-led customer interactions at scale, giving firms a unified view of conduct risk across all channels. From a Scottish tech perspective, it underlines that some of the most commercially ready AI products here are not headline-grabbing chatbots but embedded tools that help institutions satisfy regulators.
Aveni argues that adoption of agentic AI in financial services is being held back less by model performance and more by an “assurance gap” around oversight and accountability. The company says only around 2% of firms currently report having adequate AI guardrails in place, a figure Aveni attributes to industry survey data, despite growing interest in deploying AI systems that can act autonomously in consumer-facing roles. That framing lines up with a wider shift in the UK regulatory conversation, where the mode of engagement — human or machine — is increasingly treated as secondary to the outcomes experienced by customers.
The flagship new product, Agent Assure, is described as a natural extension of Aveni Detect and is designed to monitor the conduct risk of AI agents alongside human agents in a single system. Together with Agent Approve, which focuses on governing how AI agents are configured and deployed, the tools are meant to give firms a practical way to show they are treating customers fairly even when parts of the interaction are automated. For banks and insurers weighing Scottish-developed tools against global AI platforms, the promise of sector-specific guardrails built on UK regulatory expectations could be a key differentiator.
Underpinning the platform is FinLLM, Aveni’s proprietary suite of what the company describes as small language models, trained specifically on UK financial services data rather than general internet text. Aveni Labs’ own technical material places FinLLM in the seven-billion-parameter range — small relative to frontier general-purpose models, though larger than the sub-billion-parameter systems often labelled small language models in academic taxonomy. The company says this focus allows it to handle domain-specific language and regulatory nuances more reliably than generic large language models tuned after the fact, positioning Aveni within the growing category of vertical AI vendors building specialised models and compliance-ready workflows for sensitive sectors like finance.
Aveni’s Chief Scientist and co-founder, Professor Lexi Birch, holds a Personal Chair in Multilingual Natural Language Processing at the University of Edinburgh’s School of Informatics and is a Fellow of the University’s Generative AI Lab. Her role gives the company a direct link into one of Scotland’s most active AI research environments, which has produced commercial spinouts across language technology, life sciences and quantum computing. That research base matters in a market where regulators and institutional buyers are increasingly asking detailed questions about training data, model behaviour and bias mitigation rather than just headline accuracy scores.
The startup was supported into existence in 2018 by Edinburgh Innovations, the University’s commercialisation arm, and received early seed funding from Old College Capital, the University’s venture investment fund. Aveni went on to close an £11m Series A in 2024, which Edinburgh Innovations described as one of the largest in Scottish business that year, and is now based at the Edinburgh Futures Institute. The fact that Old College Capital is still close to the company several rounds later is the part of the deal structure most worth watching from a Scottish ecosystem perspective.
Old College Capital’s Head of Investment, Katharine Fox, said the firm is “helping financial services to unlock the potential of AI in a practical and meaningful way” and argued that its focus on trust and assurance “sets a benchmark for responsible AI use”. She added that the fund has been “proud to support the team” as they build momentum across the market and intends to continue backing the company through its next phase of growth. Follow-on participation from the same university fund across the seed, Series A and this round is rarer than headline raises suggest, and is the kind of signal other Scottish spinouts will read closely when they think about their own cap-table journey.
Chief executive Joseph Twigg framed the raise as an endorsement of a seven-year push to build models, domain expertise and regulatory relationships tuned to the hardest problems in AI adoption. “We have spent seven years building the models, the experience and the regulatory relationships that make us uniquely qualified to solve the hardest problem in AI adoption right now: how do you assure the conduct of an AI agent interacting with a real consumer? Agent Assure is our answer — and this investment accelerates our ability to deliver our full platform at scale,” he said. For Scotland’s fintech cluster, the test now is whether tools like this become the default layer of assurance as regulated firms move from AI pilots to production agents over the next few years.
