A CFO responsible for the books of several companies faced the same task every month: turn each entity's profit and loss statement, each in a slightly different format, into the financial models the business depends on. It was hours of careful manual entry, it was error-prone, and it grew with every entity added, which capped what one finance function could handle. New Clarity built a workflow that reads each statement in its native format, maps the line items into the standard model, validates every figure, and holds the result for the CFO's review before anything goes out. Close time fell by roughly forty percent and manual entry errors dropped to near zero.
Off-the-shelf AI tools could draft commentary or answer questions, but none could do the structured, checkable data work a financial model requires, and none could be trusted to move a number from a statement into a model without a person re-checking it. The CFO needed the entry done, the checking built in, and the final decision left with a person.
A custom connector that gives the model structured, permissioned access to the finance workflow, built on an open standard for connecting models to tools. It reads each P&L in its native format, maps the line items into the standard model, and validates each value against the expected pattern, flagging anything out of range rather than accepting it. The CFO then works with the draft in plain language: adjust an assumption, drill into a variance, or produce commentary for the board or the owners. Because the connector is standard rather than bespoke, one connector serves every entity without per-entity configuration, keeps working as the underlying models improve, and extends to other parts of the finance function without starting over.
The workflow runs in the firm's own environment. Every figure is validated on the way in, anomalies are flagged for a person, and the CFO confirms the model before it goes out. The record shows what was read, what was mapped and who approved it.
Month-end close time fell by roughly forty percent, manual entry errors dropped to near zero, and the finance function took on more entities without adding hours. The same design applies to any group that consolidates several entities, including a fund's holding and blocker entities and the operating partner reports that arrive in a different format from each partner.