Latest top stories
Technology

Vena’s Planning Agent: Can AI Rescue the Struggles of Finance Teams?

25 September 2025

The spreadsheets, the late nights, the endless reconciliation — if you’re in finance, chances are this portrait is familiar. The promise of artificial intelligence in such a world is seductive: faster forecasting, fewer errors, more headspace for strategy. But the stakes in finance are high. It’s not just about speed; it’s about trust, accuracy, auditability.

This month, Vena — a player quietly entrenched in the world of financial planning and analysis (FP&A) — launched its Planning Agent, an AI assistant that promises to dramatically accelerate budgeting, forecasting and scenario analysis. It is part of a broader suite of agents built on Vena’s existing Excel-centric platform. We spoke with Vena’s CTO, Hugh Cumming, to understand what this means in practice.

According to Cumming, early Planning Agent customers, in private preview, saw their budgeting cycles shrink by over 60%. That is a headline number. The tech also recently won the Highest Business Impact Award at Vista Equity’s North America Agentic AI Hackathon. 

These gains suggest not just incremental improvement, but a possible shift: from planning as a periodic slog to planning as a living, reactive capability.

A quiet evolution: how Vena built its footing

Vena was founded in Toronto in 2011 by finance professionals who believed that better planning should lead to better performance. From its early days, its mission has been to empower businesses with FP&A tools that mesh with existing workflows. Today, Vena describes itself as an AI-powered Complete Planning platform, deeply woven into the Microsoft ecosystem: Excel, Power BI, Teams and more. 

That “Excel-first” stance is deliberate. Many finance teams resist being forced into entirely new interfaces. Vena has instead tried to inject governance, data workflows, version control, audit trail and scalable planning logic under the familiar surface of spreadsheets. It is listed on Microsoft’s AppSource as an “AI-powered, Complete Planning Platform purpose-built to amplify your investment in the Microsoft tools you know and love.” 

According to business intelligence sources, Vena serves over 1,800 customers globally. It has been recognized repeatedly in FP&A and enterprise planning vendor rankings. Its approach is not flashy, but methodical: embed deeply, win trust, and gradually layer intelligence.

And intelligence is precisely what the firm now aims to bring to planning.

The challenge: time, complexity and the limits of spreadsheets

Cumming frames the core dilemma plainly: finance teams are overwhelmed with administrative burden. Most of their time is swallowed by data gathering, formatting, and mundane workflows — leaving little capacity for strategic insight. He says:

“Eighty-two percent of CFOs have said that their responsibilities have only grown. Gathering data and administrative processes take up around 75% of an FP&A team’s load, leaving them with little time to tackle big strategic planning, forecasting, and analysis.”

That is a painful admission for many organisations. Budgets still take weeks. Forecast cycles lag. By the time results are in, circumstances have often shifted.

That is where Vena’s AI agents are meant to step in: automating data collection, distilling insights, and presenting usable outputs (in Excel, in PowerPoint). In effect, the agent framework is intended to let finance teams reclaim time to think, model, and guide.

What the Planning Agent can (and cannot) do — and what makes it distinct

In Vena’s architecture, the Planning Agent is not a standalone toy, but part of a concerted agentic framework (alongside Reporting, Analytics, Query). In conjunction with their Reporting and Analytics Agents, the Vena Planning Agent covers the entire financial life cycle.

Key features include:

  • Driver-based planning: Rather than blindly increasing or flat-lining based on prior periods, the agent works with business levers (sales, costs, growth drivers) to produce forecasts grounded in logic.

  • Predictive forecasting: Historical data is augmented with external signals (macroeconomic, trend indicators), which may counteract inherent bias or overfitting.

  • Scenario modeling: With a few prompts, users can test hypothetical futures (“if demand falls 10%,” “if input costs rise 4%”) and see results contrasted with baseline plans.

  • Excel-native interface: The agent doesn’t demand you leave your familiar spreadsheet world; it works within the same interface, bridging human comfort with algorithmic speed.

 

Strategic value: when speed becomes strategic

The value of speed is not simple. If forecasts are delivered faster but less accurate, the benefit may be illusory. But if speed comes without losing confidence, then planning becomes tactical, continuous, and levered.

Cumming illustrates with a practical scenario:

“Say you’re a manufacturer importing a lot of materials or goods from multiple countries. As trade deals and tariffs go into effect, you have to constantly re-forecast what your cost structure will end up being. The Planning Agent has the information and knowledge base to help an organization respond to those sudden moves without taking days.”

In volatile geographies, with supply disruptions, cost swings or regulatory shocks, a planning tool that can re-run assumptions swiftly becomes more than convenience — it becomes power.

Because Vena already sits in many finance stacks, its incremental value is in plugging into existing workflows rather than forcing major changes. This may make adoption less painful than jumps to entirely new systems. Moreover, integration with Microsoft — a platform many firms already use — helps it embed itself without demanding wholesale architectural change.

The trust gap: how finance must govern AI

Even the most elegant AI must convince humans that its outputs are reliable, transparent and controllable.

Cumming recognises this explicitly. Finance professionals rightly ask:

“Is this data going to be accurate? Is my data protected? Do I retain control over my data?”

To address these, Vena has built multiple guardrails:

  • Customer data stays within their own Vena Tenant, hosted on Azure, and is not shared to train publicly accessible models.

  • Their CubeFLEX™ system carries forward existing security configurations (data access, application permissions) into the AI layer.

  • A Prompt Flow architecture logs all questions, responses and allows rollback and inspection of prompt logic.

  • Administrators can review any interaction, flag and correct responses, and shape training hints or rules.

  • Controls exist over exactly what data the AI can access and which prompts it can run.

  • AI is advisory, not prescriptive: final decisions remain in humans’ hands. As Cumming puts it:

“Vena Copilot and the agents can also be trained and further tailored to specific business needs." Cumming emphasised. "Admins have complete control, which adds the benefit of improved accuracy. Vena Copilot and the agents put information and insights into the hands of human decision makers, so they choose what to do with what AI gives them.”

These guardrails are essential, but their utility will be tested in edge cases: unexpected entries, restatements, audit challenges, regulatory ambiguity. That is where true trust is earned or lost.

What lies ahead — adoption, risk and evolution

Cumming envisions finance not as a back-office function, but as an operation’s centre:

“From our point of view, the next step for AI in finance is about using AI’s automation capabilities to turn finance into an operational organizer for an entire business. We plan to continue delivering tangible value, by combining AI autonomy with human strategic oversight.”

The Planning Agent is now in closed beta with select customers, It is expected to become generally available in November, with the Query Agent (for natural language queries) later in the year. Beyond that, additional agents are slated for 2026.