Curated by Rob Saunders, WhoFiled. Interested in launching an industry intelligence report for your audience? Contact rob@whofiled.com.
The big theme: now that agents are live in regulated workflows, the budget is shifting from the agent to the proof. Verifying what an agent was authorized to do, and recording what it did, is becoming its own line item, the one the platforms will buy rather than build. An auditor gains more credibility when it’s independent of the auditee.
Aveni, an Edinburgh compliance-software firm, raised £12 million led by PXN Ventures, with Puma Growth Partners, Lloyds Banking Group, Nationwide, and Scottish Enterprise returning. The money funds two products, Agent Assure and Agent Approve, that judge the conduct risk of AI agents touching clients. Aveni cites only 2 percent of firms reporting adequate guardrails for the agents they are already deploying. Lloyds and Nationwide are deploying the agents and bankrolling the company built to police them, in the same round. A Grant Thornton survey found 78 percent of executives lack strong confidence they could pass an independent AI governance audit within 90 days, and a separate American Arbitration Association benchmark put the share of financial firms very confident they can produce auditable evidence of AI governance at 21 percent. Capability shipped ahead of the ability to defend it.
Regulation is now setting the clock. In Japan, Mizuho and NEC began trials this month of an authentication layer they call Know Your Agent, verifying an agent’s identity, authority, and consent to act, with a record of every action. Banks have always had to verify the identity of customers; now they verify the identity of its software. The same logic arrives in the US through the calendar: the SEC’s amended Names Rule reaches its first compliance date for larger fund groups this month, forcing funds to show that at least 80 percent of assets match what the fund’s name implies. The rule was not written for an AI world, but its deadline sets a hard date on the evidence market.
Notable investment tech raises
AlphaSense raised $350 million co-led by Vitruvian Partners, Accenture Ventures, and J.P. Morgan Asset Management at a $7.5 billion valuation, with Accenture also signing on as a strategic channel partner. The AI market intelligence platform sells a corpus of over 500 million filings, transcripts, and expert interviews. It crossed $600 million in annual recurring revenue, and the capital-markets CFO hired alongside the round signals an IPO.
(Roughly $640 million this week flowed to those that verify, govern, secure, or feed agents against about $75 million to the two selling agents that do the work outright, eight to one for proof over the agent itself.)
Gradient Labs extended its Series A to $26 million for vertical AI agents that each handle one regulated banking function, with FCA Consumer Duty and EU AI Act tests built in. Bayshore raised $8 million in seed to turn regulations and internal policies into machine-readable code that agents execute.
Eventus, the trade-surveillance provider taken majority by Terminus Capital in February, named a new chief technology officer and chief customer officer this week as it scales its Frank AI surveillance platform. They hired a customer officer, not a sales one: the tool is built, and the desk that once watched traders is now asked to watch the agents.
On the radar
Policing the agent is now its own venture line. Gray Swan raised $40 million to red-team frontier models and has been cited in eleven recent model system cards. Willow emerged from stealth with $7 million for a governance layer over agent actions, citing a 2026 survey in which 65% of companies reported an agent-related security incident in the past year, and Offroad raised $7 million for identity security spanning human, machine, and AI identities. The data layer repriced the same way: Coralogix raised $200 million as agent-generated telemetry overwhelms legacy monitoring, the audit trail becoming its own platform.
Consolidation is now the buyer’s stated strategy. London-based Marathon Asset Management retired its legacy systems for a single front-to-back platform, joining Pictet and AllianceBernstein , and SimCorp’s 2026 InvestOps Report found consolidation is the top priority for 58 percent of managers. Epiris carved out Winterflood Business Services, and Repay closed its $372 million purchase of Kubra. AI was supposed to make the fragmented stack talk to itself. Instead it talked buyers into throwing the stack out. Amplify, an RIA operating platform now running 655 advisers and $24.5 billion, added a Wealthbox integration. Saris raised $28.8 million to automate bank back offices and Pace raised $46 million to run insurance workflows with agents, the same bet two verticals over. Integration, once the line item buyers paid most to avoid, is now the free hook, and the workflow it captures is the product.
Robinhood escalated from courting advisers to subsidizing them with forgivable loans for registered investment advisers, buying the adviser relationship before the agent tools meant to run through that pipe have shipped. And Ramp raised $750 million at a $44 billion valuation, naming token spend management for AI agents as a growth driver. The meter is going in before the agents to meter exist at scale.
One quieter through-line: Aveni, Epiris, and Marathon are all British, a forward read on where US compliance budgets head next.
The takeaway for investors
Do one thing this week: pull every AI tool already running in your shop and ask the vendor for the audit trail, not the output. If they cannot show what the agent read and why it acted, you own the conduct risk. The firms that can answer deploy at scale; the rest are running a pilot they cannot defend.
Data sourced from SEC Form D filings, developer activity, and alternative signal tracking by Rob Saunders at WhoFiled. Specific developments are linked to primary sources inline above. Rob Saunders is exclusively responsible for its accuracy. If you have any feedback, please contact us.