Finance teams did not build spreadsheet-heavy workflows because they love spreadsheets. They built them because the systems they paid millions of dollars for could not answer the questions they needed answered.
For the past two decades, enterprise finance software has become very good at storing data. ERPs, billing systems, financial planning platforms and financial infrastructure tools serve as systems of record for nearly every financial transaction inside a business.
Yet much of the actual work of finance happens somewhere else.
When executives ask why revenue missed the forecast, which customers drove churn, why margins compressed or whether hiring plans remain on track, finance teams rarely answer those questions directly in the ERP. They export data into spreadsheets. They pull information into Snowflake or BigQuery. They build dashboards in Tableau, Looker or Power BI. They create financial models in Excel that sit outside the systems where the underlying data originated.
Finance software became the source of truth while spreadsheets, warehouses and BI tools became the place where that truth got interpreted. Over time, finance organizations built an entire operating layer on top of their systems of record, much of it held together by institutional knowledge rather than formally governed definitions.
The Last-Mile Finance Stack
Most finance leaders can probably sketch this architecture from memory.
Financial data originates inside operational systems. Teams move that data into warehouses, spreadsheets, dashboards and reporting environments where they can manipulate it, combine it and answer business questions.
Businesses rarely operate exactly the way software vendors expect them to. Revenue recognition gets customized. Forecasting models evolve. Departments develop company-specific metrics. Board reporting requirements change. Acquisitions introduce new definitions and processes.
Finance teams responded by building their own layer on top of the systems of record. Spreadsheets offered flexibility that packaged software could not. Warehouses made it easier to combine information from multiple systems. BI tools made it possible to distribute reporting across the organization.
Every layer solved a problem, and every layer also introduced opportunities for inconsistency.
Over time, financial logic became distributed across spreadsheets, dashboards, warehouses and planning systems. Organizations adapted by creating processes to reconcile information and keep teams aligned, even when the underlying systems were not.
That reconciliation work exists because the information needed to run the business became fragmented across dozens of tools and workflows.
Why AI Changes the Equation
AI agents introduce a different operational model. Instead of pulling information out of systems and moving it through a chain of spreadsheets, dashboards, presentations and emails, an agent can query the underlying systems directly, perform analysis and return an answer.
That distinction matters because much of the last-mile finance stack exists to move information from one place to another. Analysts extract data from one system, transform it in another, visualize it in a third and distribute it through a fourth.
When software can access underlying records directly and perform analysis on demand, some of those primarily manual workflows stop looking like permanent requirements and start looking like workarounds that accumulated over time.
Finance teams will still use spreadsheets, warehouses and dashboards. What may change is how often information needs to move between them before someone can answer a question.
The bigger shift is that analysis begins to happen closer to the systems of record themselves. In an environment with strong governance, finance leaders no longer need to know which system contains the answer. They can ask a question and trust that software will retrieve the relevant data, apply consistent definitions and return a reliable result.
The Problem Was Never Spreadsheets
Spreadsheets have become a convenient villain in finance discussions, but spreadsheets were never the root problem.
Finance teams use spreadsheets because they provide flexibility and transparently auditable logic. Analysts can test assumptions, build models, investigate anomalies and answer questions that packaged software cannot.
The real problem is not spreadsheets. It is the lack of definition governance across finance, sales, operations and leadership.
Revenue, churn, ARR, customer counts and forecast assumptions often exist in multiple systems with different interpretations, ownership and business logic. Over time, those differences become embedded in dashboards, planning models, warehouse transformations and spreadsheets.
Eventually, nobody can explain where every number originated without tracing the entire chain manually.
AI does not solve that problem. In many cases, it exposes it.
An agent cannot reliably answer a financial question if finance, sales, operations and executive leadership all maintain different definitions.
Governance Moves to the Center
For years, finance teams compensated for inconsistent systems through human effort.
Analysts knew which spreadsheet contained the approved metric. Finance managers understood which report represented the official number. Institutional knowledge filled the gaps between systems.
Agents do not operate on institutional knowledge. They operate on context: written down definitions, permissions and data structures.
If finance, sales and operations define revenue differently, automation does not resolve the disagreement. It simply retrieves an answer faster than before.
Finance teams have spent years solving data collection problems. Agents can automate much of that work. What they cannot automate is governance.
Someone still has to decide what counts as revenue, how churn is measured, which forecast is authoritative and which version of a metric belongs in a board deck.
Those questions become more important as automation becomes more capable.
What CFOs Should Evaluate
Many finance software evaluations still revolve around dashboards, reporting capabilities and user experience.
Those factors matter, but finance leaders increasingly need to evaluate a different set of questions as well. For example:
- Can the system expose data cleanly?
- Can it support consistent governed definitions across departments?
- Can it explain how conclusions were generated?
- Can it provide the controls necessary for financial oversight?
- Can it serve as a reliable source of truth for both humans and machines?
These questions matter because agents are beginning to plan analysis instead of just executing what the human planned.
Finance built the last-mile stack because systems of record alone could not answer the questions the business needed answered. Warehouses, spreadsheets, dashboards and planning models filled that gap.
Agents attack the same problem from a different direction. Instead of moving systems, agents conduct the analysis across multiple systems using your governed definitions.
Whether that eliminates large portions of the last-mile stack remains to be seen. What is already clear is that automation increases the cost of inconsistent definitions. A human analyst can usually recognize when two reports disagree. An agent will confidently return whichever answer the underlying systems provide.
That puts governance, ownership and financial definitions at the center of the conversation. Finance teams have always been responsible for those things. AI makes the consequences of getting them wrong much harder to ignore.
