What's new in Vena: the Spring '26 release

What is the Vena MCP server?

Status: Closed beta

AI assistants inside Excel are useful, but they share one limit: they only know what's in the sheet in front of them. Most of your data is far too large to load there, which is often why you use Vena in the first place.

The MCP server (MCP stands for Model Context Protocol) connects whichever model you're using, Claude or ChatGPT, to your governed Vena data layer. The AI can then work against all of your Vena data rather than a single sheet: pull it in, build a model from it, run analysis, etc.

This is where it gets useful. A general-purpose model with no context will happily build you a revenue template, but it's working from placeholders and guesses about your business. Point that same model at your Vena data through MCP and it builds from the real thing: your actual product hierarchy, your scenarios, your periods and rollups, structured the way your model already is.

And because you're still working with a general AI assistant, you keep everything else it brings with it. You can teach it your company's formatting through Claude Skills, so every model it generates comes out in your own fonts, colors, and styles rather than its defaults. You set that up once from your existing templates and share it across the team. The same goes for the other tools and connections those assistants plug into; your governed Vena data simply becomes one more source the AI can reason over, alongside the rest of how your team already works. That combination is what makes this more than another chat box in Excel.

It's in closed beta for now, so it isn't switched on for everyone yet. If it's something you'd want to pilot, it's worth flagging early.

What are operational tables, and how do they connect to Power BI?

Status: Open beta

The Vena cube is built for rollups and fast calculation, not for holding large volumes of transactional detail. Operational tables change that. They let you store relational data — journal entries, line-level headcount and salary records, and the like — directly in Vena, and read and write to it from the Excel add-in. The cube stays lean as the rollup layer, while the heavier detail sits in tables built to handle enterprise data volumes.

You can combine that operational data with your CubeFLEX models, so a plan can draw on far more context than before, all without leaving a governed environment. The usual reason teams keep this kind of detail in side spreadsheets is that it doesn't fit neatly in the cube; operational tables remove that reason, which means less data living outside any system of record.

The same data also feeds Power BI. A new Power BI Desktop integration pushes your operational table data straight into Power BI, so you can build dashboards and visual analytics on the exact figures your team is planning against. This means no separate export, no overnight refresh, and no gap between the numbers you plan with and the ones you report on, even down to the transactional level. For teams that plan in Vena but report in Power BI, it keeps both sides working from one governed source rather than two copies that drift apart.

What's new in Vena Copilot?

Status: Generally available

Copilot has had an accuracy pass in this release: a better read on your data model and more reliable calculations behind its answers. You can also see how it interpreted a request and adjust that interpretation before acting on the result. Separately, the reporting agent is now included for every Vena Copilot user in the ad hoc tool at no additional cost.

Which Vena agents are now generally available?

A few capabilities that were still in beta at the winter release have since graduated. If you read our winter write-up, these are the same agents, now out of beta:

• Planning agent: build driver-based plans from a plain-language prompt, directly in Excel, without setting up formulas and mappings by hand.
• Query agent: turn a plain-English request into ready-to-run MQL, with no syntax knowledge required.
• Reporting agent: describe the report you want and have it pre-built when you open the ad hoc tool, rather than starting from a blank sheet.

Frequently asked questions

Is the Vena MCP server available now?

Not for everyone yet. The MCP server is in closed beta, so it isn't switched on by default. If you'd like to pilot it, it's worth flagging early so you can be included while the beta runs.

Which AI assistants work with Vena through the MCP server?

The MCP server connects general-purpose models — Claude and ChatGPT — to your governed Vena data layer. Whichever assistant you use can then work against all of your Vena data, not just the sheet in front of it.

What's the difference between the Vena cube and operational tables?

The cube is built for rollups and fast calculation. Operational tables are built for relational, transactional detail — journal entries, line-level headcount and salary records, and the like. The cube stays lean as the rollup layer, the detail sits in the tables, and you can combine the two in your CubeFLEX models.

What this means for your team

This release doesn't hang on a single headline feature. Most of it works toward the same end: closing the distance between your governed data and the place you actually do the work. The agents that were experiments a few months ago are now generally available, and the bigger bets, the MCP server and operational tables, are in beta and ready to test.

If you'd like to talk through which of these are worth piloting first, feel free to reach out. We're happy to spend some time on it.