CUBE Connect 2026: what we heard

AI is already changing how compliance and risk teams work. At CUBE Connect 2026, in London and New York, the conversation was about making that change something every team, and every regulator, can trust.

The numbers set the scene. 59% of firms are embedding AI, 57% distrust it, and only 34% are confident they could demonstrate AI governance to a regulator. That gap between adoption and confidence is what CUBE is built to close: an AI workflow and a trust standard resting on good data. Regulatory data infrastructure, regulatory intelligence and orchestration work together, so every data asset carries end-to-end traceability and your teams can show how an outcome came to be.

From strategy to the screen

The live demo of the unified RegPlatform™ showed what that looks like in practice. Deep Analysis filters by jurisdiction and sets the depth of the answer. A compliance map links obligations to policies, products and controls. Global AI Search answers one question across jurisdictions, and an orchestration calendar runs compliance and risk workflows end to end, with agentic Coworkers taking on the repetitive work along the way.

Platforms are only half the story, though. The same regulatory data can power the Regtech, Legaltech, Fintech and Bigtech worlds. The common thread was curation: reliable data comes from how it is built, and "fiduciary-grade AI" means free of hallucinations and mistakes.

Customer case studies

The customer sessions showed how firms are leveraging CUBE Coworkers - the purpose-built AI agents that sit inside RegPlatform™. They work from the regulatory intelligence, inventory and firm-specific context customer teams already rely on, and keep every output source-linked and auditable.

Start with the time it takes to reach an answer. One customer described its compliance work as a very manual process before AI, and said outside help would have meant a multi-week project, whereas now a regulation already in CUBE's inventory can be mapped in minutes.

Then there is the noise. One customer described fine-tuning its profile because reducing the volume reduces the noise, focusing on the relevance of what goes into the funnel and separating high risk from low risk. Another customer asked the question every team asks, namely how to be sure nothing is being missed, and pointed to the audit trail as the answer. Priorities Coworker is designed for exactly this: it ranks the highest-impact RegInsights by certainty and applicability, tailored to your regulatory profile, with a full audit trail of assessment decisions.

All customers kept people at the heart their success stories; CUBE's Coworkers are built the same way, helping teams move faster without completely changing how they work.

AI and the world of governance

Governance starts at the very beginning, and the Governing the machine panel was clear that every step of the process needs it, including who is accountable for each decision. Accuracy, completeness and reliability are the critical elements, and 'human in the loop' (or expert in the loop) cannot become a box-ticking exercise. When something goes wrong, firms need to show a regulator their control environment and how they got there.

The hard questions sit around agents. If ten people have touched a model, who is accountable? Agent identity, lineage and traceability were framed as contractual versus regulatory protection. Something can crash in real time, so continuous monitoring replaces older compliance processes, which puts more pressure on compliance and risk. The first line still owns primary risk, particularly customer risk when AI faces customers. Sub-agents add another layer: agents can create them, and engaging with the primary agent will not reveal them, so the room discussed tools to identify the layers. And because AI does not understand your business the way your staff do, the question to keep asking is how AI helps you run it better.

Experts in the loop

That focus on accountability leads straight to people. Across both events, speakers pushed past "human in the loop" to something more precise: experts in the loop. Human oversight is a requirement today and may relax as trust in the data grows, but accountability stays. When an agent gets it wrong, someone must answer for it. As one speaker put it, "You can't say it wasn't me, the AI agent did it."

In practice, that means delegating the work, not the responsibility. Panel discussions pointed to a single control tower and front door for AI, observability across every agent, and a watch for drift as capabilities grow. It also means investing in people: SMEs, upskilling and AI literacy, including a clear view of what AI can and cannot do. Some suggested the operating model itself could shift from three lines to a single, proactive line. And on regulators, the advice was direct: don't wait, engage now.

One language across the industry

Consistency ran through every session. The goal is not perfection but one model that stays nimble enough for local nuance, held together by a common taxonomy across risk, compliance and obligations, so every line tells the same story to regulators. If the data going in is poor, so is the outcome, which is why data quality came up again and again.

The industry collaboration initiative puts that into practice. Every institution is solving the same problems, alone, and the initiative is built on trust by design, with non-attribution at its core. It aims to deliver a shared language, open industry standards and non-identifiable benchmarks. So far: 12 public statements, 4 working groups and 10 members.

What comes next

Experts stay accountable. A shared language makes both work at scale. This is what the future of regulatory change management looks like, and there's more ahead. To see the unified RegPlatform™ in action, let's talk.

Trust CUBE.