What shipping finance-grade intelligence actually looks like

In the first three editions of Intelligence inside, Elias and Kevin explained the foundations: why the bar for AI in finance and tax is higher than in almost any other software category, how we built the Intelligence Core and its trust architecture, and how the Data Platform and semantic layer give our agents a governed, grounded single source of truth to reason over.

That bar, trustworthy on day one, traceable on day two, defensible in front of an auditor on day three, is not just a design principle. It is the filter every product decision in my team goes through.

This edition is about what that filter produced: a year of intelligent features already live across our CFO Solution Platform. Not announced, not in beta, not on the roadmap. Shipped, in production, and in use.

 

The rule we apply to every feature: does it earn the right to be there? 

Kevin described the division of labor clearly: deterministic logic does the calculation, intelligent reasoning does the interpretation. That distinction is not just a technical choice. It is a product discipline that we apply before a feature gets built.

The question I ask before any intelligent capability gets close to a customer is simple: does this deserve to be in a CFO's workflow? Not “is it technically impressive?” Not “does it work in a demo?” Does it solve a specific, named problem in the way a finance or tax professional would want it solved, with full visibility, full control, and no surprises?

Every feature I will describe below passed that test. Each one takes on the repetitive, judgment-heavy, or expertise-dependent work that slows teams down, and hands the result back to the professionals for review.

The intelligence does the work. A human approves it. That is not a constraint, it is the design. 

What that looks like in practice varies by solution. But the principle is consistent across all of them.

Smarter models, faster: intelligence in financial planning

Building a financial model from scratch has always been a specialist's job. You need to know which variables matter, how to structure the formulas, and how to connect the model to the actuals underneath it. For most teams, that setup alone takes days, and is dependent on one or two people who know the model architecture well.

We started shipping intelligent modeling features in xP&A in May 2025. Not as a single big launch, but as a deliberate sequence.

 

From AI-suggested variables to fixing errors in one click

First came AI-generated variable suggestions: the system analyzes what is already in the model and proposes what comes next, including names, formulas, and hardcoded assumptions. Finance teams accept or reject suggestions individually. The blank-canvas problem, the hours spent figuring out how to start, goes away.

Then AI-assisted forecasting: instead of manually defining a forecast methodology, the system analyses historical data and generates the most likely projection. A starting point that used to take hours arrives in seconds.

Then Prompt AI, in July 2025: describe what you want to calculate in plain language, and the system proposes the variables and structures needed to produce it. And most recently, in April 2026, Fix with AI: hover over an error, click once, review the proposed correction. What previously required debugging expertise now takes a single interaction.

Taken together, these features reduce the time to build and maintain a financial model significantly. Each one hands the result back to the finance professional for review. The model stays theirs. These are the early innovations in xP&A. The next step is not a feature, it is a workflow.

Intelligence in ESG Reporting: Automated emission factor mapping

Greenhouse gas (GHG) footprint calculation is a task finance teams are increasingly responsible for but rarely trained to do. Matching emission sources to the right emission factors across all scopes requires either specialist knowledge or expensive external support. Most teams have neither.

What we built in ESG Reporting removes that dependency. Teams upload their consumption data, and the system maps each entry to the correct emission factor automatically across scopes 1, 2 (location- and market-based), and 3, reducing the required GHG expertise. Every mapping comes with a confidence score. The team reviews, overrides where needed, and moves on.

The result is generating GHG footprints five times faster than with manual emission factor mapping. For teams under CSRD pressure, that’s the difference between meeting a deadline and missing it.

From blank page to board-ready annual reports

Annual report creation sits at the end of the hardest part of the finance calendar. By the time teams reach the drafting stage, they are already under deadline pressure, staring at dozens of disclosure requirements that need to be written, reviewed, translated, and aggregated across entities.

We introduced intelligent writing capabilities in Disclosure Management in January 2025.

 

Drafting and optimizing texts with intelligence

Two functions came first: text optimization (adjusting tone, shortening, expanding, translating, summarizing), and taxonomy-compliant text drafting. For any ESRS or IFRS disclosure label, the system generates a compliant draft. Teams provide company-specific context; the system incorporates it. The report language and layout are applied automatically.

ESRS-compliant group reports

By April 2025, we added group-level ESRS report generation: for organizations with multiple reporting entities, the system consolidates subsidiaries' questionnaire responses and generates a final, ESRS-compliant sustainability report at the group level. Teams select which inputs to include, add supplementary information, and can regenerate until the output meets their requirements. What used to take hours of manual aggregation across entities now happens in a single guided process in minutes.

 

Regulatory filing without the specialist bottleneck: XBRL Tagger

XBRL tagging is one of those tasks that sounds technical because it is. Since 2020, listed EU companies have been required to file in the Inline XBRL (iXBRL) format. The tagging process of matching report elements to taxonomy concepts across thousands of line items has historically required specialist consultants. For most mid-market finance teams, that meant meaningful cost and external dependency every filing cycle.

In December 2025, Lucanet launched the Tagger Agent, becoming the first in the market to ship a comprehensive AI agent for XBRL reporting. Teams upload their annual report and the Tagger Agent handles narrative tagging, numeric table tagging, and calculation generation automatically. No implementation project, no IT involvement, and no need for consultant support. The Tagger immediately loads any valid taxonomy, including ESEF, local GAAP variants such as Dutch GAAP, European-level mandates like pCbCR, and sector-specific mandates like OCW. It then produces iXBRL output that meets current regulatory requirements.

Once the mapping of taxonomy concepts is established for a filing, it carries forward automatically. The specialist bottleneck does not return.

What a year of shipping has taught us

The features described above did not arrive as a bundle. They arrived as a sequence, each one solving a specific problem in a specific workflow, each one designed so the finance professional stays in control of the output.

Kevin previously shared a useful framing: a new hire with PhD-level intelligence, but day one on the job. Raw capability is necessary, but not sufficient. What makes it useful is grounding it in the specific context of your business: your consolidation structure, your reporting conventions, your data. That is exactly what these features do. They bring intelligent reasoning to the specific tasks your team does every day, grounded in your data, governed by the Intelligence Core, and always including the human in the loop.

That is what "simply intelligent" means to us in 2026. Not intelligence for its own sake, but intelligence that earns its place by being genuinely useful, genuinely trustworthy, and genuinely under the control of the people accountable for the numbers.

And this is not where it stops. In the coming months, planning model templates will be generated automatically from actuals, giving finance teams a structured starting point without the setup work. Excel imports for sustainability questionnaires will be mapped to the right questions automatically, removing manual data entry ahead of reporting deadlines. And arriving this summer: automatic lease contract data extraction will turn what was previously a manual, document-by-document task into an automated step.

Every feature already shipped is evidence that the architecture works. What comes next is built on the same foundation.

 

Ready to revolutionize your financial workflows?

Watch our on-demand webinar to see how Lucanet empowers CFOs and their teams with AI-driven automation and real-time insights.

 

Learn more

Related posts:

Intelligence inside #1: Simply intelligent in the agentic age

Intelligence inside #2: Building Lucanet’s trust architecture

Intelligence inside #3: Lucanet’s Data Platform and the semantic layer