CryspIQ 3.3.0 puts governed enterprise data inside Microsoft Copilot

CryspIQ shipped version 3.3.0 of its data platform on 12 November 2026, and the headline feature is access to governed enterprise data directly inside Microsoft Teams and Microsoft 365 Copilot. The company positions the release as an answer to a question that has dogged enterprise AI since chatbots arrived at work: can you trust what the assistant tells you when the data underneath it is shaky?

Dan Peacock, Co-Founder and Chief Executive Officer at CryspIQ, framed the problem bluntly. "Everyone wants to use AI on their own data, but the real question is whether you can trust what it tells you," Peacock said. "That trust is earned in the data layer, long before anyone types a question. With v3.3.0, we've focused on getting those foundations right, so people can act on the answers with confidence."

The stakes here are practical rather than theoretical. Companies rolling out Copilot and similar assistants keep running into the same wall: the tool is only as reliable as the records behind it. CryspIQ's bet is that governance work has to happen before a question is ever typed.

How does the Teams and Copilot integration work?

Responses in Teams and Copilot run under each user's Microsoft sign-in, and access is limited by Microsoft Entra ID security groups. That means an employee's permissions follow them into the AI conversation. If a security group bars them from a dataset, the assistant won't surface it.

Copilot reads directly from CryspIQ. Users don't need to build and maintain a separate data source just for Microsoft's assistant — a maintenance burden that has tripped up many enterprise AI deployments. Each answer also displays the underlying evidence, so a person can check where a number or claim actually came from before acting on it.

That evidence trail matters for regulated industries and for any organisation where an untraceable AI answer is a liability rather than a convenience. CryspIQ is explicitly targeting three groups with the release:

• Executives and managers who want direct answers inside Teams and Copilot
• Data administrators responsible for source-to-target mappings
• Cyber and IT teams managing data access and separation between organisations

What changes in data preparation?

Version 3.3.0 focuses on preparation and governance before information becomes queryable. The goal, according to the company, is shortening the path from a source file to a trusted data model while catching data quality issues earlier in the process.

The centerpiece is a guided setup process. Lumen, CryspIQ's own AI assistant, proposes how a source file should map into the platform's data model. Users review those suggestions, preview the result, and save the mapping. On the mapping page, high-confidence suggestions are applied automatically, leaving humans to confirm only the uncertain cases.

Data stewards also get more direct control. The mapping page now lets them map fields themselves, reducing reliance on administrators for routine tasks. It's a small change with real workflow consequences — the people closest to the data no longer wait in a queue for IT.

Can the platform fix its own data problems?

Partly, yes. The release introduces what CryspIQ calls a self-healing parking lot. In practice, business records sometimes get parked because related context records haven't arrived yet — a shipment without its order, say. Under 3.3.0, those parked records load automatically once the missing records show up. No manual reconciliation required.

Data quality detection also moves earlier, to the mapping stage itself. Records that fail checks are held for the relevant data steward, together with the reason for failure. Instead of discovering a broken dataset downstream, stewards see the problem at the point of entry.

What's new in governance and reporting?

Governance controls get a broad update. Lumen's answers operate under each user's individual permissions, and an organisation's data, mappings, lookups, and settings stay separate from those of other customers — a multi-tenancy guarantee aimed squarely at the cyber and IT teams CryspIQ wants to win over.

Delegated role management arrives for data administrators, within limits each organisation sets. A new Company Profile screen lets an organisation place its own logo in the platform's navigation — a small branding touch, but one that signals whose data environment a user is working in.

Data quality reporting has been rebuilt around drill-down. Every count on the Business Quality and Steward dashboards can now open the exact records behind it. Each issue links to its source and its assigned data steward, so accountability has a name attached. Dashboard filters persist in shared links too, which means the view a colleague sends you is the view you actually see.

Why does the underlying data model matter?

CryspIQ's broader approach rests on a single business-event data model that structures records from multiple systems, so they can be queried consistently over time. That architectural choice is what makes the Copilot integration viable: the assistant isn't querying a tangle of disconnected sources, but one governed model.

Vaughan Nothnagel, Co-Founder and Chief Technology Officer at CryspIQ, laid out the design philosophy: "What drove v3.3.0 was a simple idea: businesses should be able to get an answer they can trust, quickly, from the data they already have. That means getting new data in faster, catching problems before they cause trouble, and making sure that when someone asks a question, the answer shows its evidence and respects what that person is allowed to see. Businesses gain confidence, which is increasingly important and more challenging to realise. Teams spend less time chasing data issues and waiting on reports, and leaders can act on answers knowing where they came from."

What happens next?

The release is live now. Its significance extends past one vendor's changelog: Microsoft has pushed Copilot into millions of Microsoft 365 seats, and the gap between an assistant's fluency and its factual grounding remains the central enterprise AI problem. Vendors that can guarantee permission-aware, evidence-backed answers — rather than plausible guesses — are positioned to capture the organisations that have paused their rollouts over exactly those concerns. CryspIQ's 3.3.0 is a bid to be one of them, with the bet placed firmly in the data layer.