AI accounting software for firms: build a realistic capacity scenario

01Start with one workflow and one unit of work

A capacity estimate becomes useful when everyone knows what it counts. For document entry, the unit could be an invoice or receipt. For bank reconciliation, it could be a statement line. For management reporting, use a completed reporting pack. Keep those workflows separate: a bank line and an invoice can represent the same underlying transaction, so adding both volumes without checking the process can count work twice.

Pick a representative month and record the volume, preparation time, review time and follow-up time. Include documents that arrived late or needed correction. A neatly prepared demonstration file can help you understand the interface, but it cannot establish your firm's likely capacity on its own.

02Include the work that remains with people

An automated draft still needs the appropriate accounting treatment, source support and review. Some items need context that software does not have: an unfamiliar supplier, a mixed personal and business purchase, or a payment with no supporting invoice. Estimate routine review separately from extra exception handling so both remain visible.

Ask reviewers to log the time spent correcting drafts and checking exceptions during the pilot. If preparation gets faster but review becomes harder, the apparent saving can disappear. Also record work that happens outside the portal, including obtaining explanations from clients and resolving missing records.

• Current effort: document volume multiplied by measured minutes per document.
• Scenario effort: preparation plus routine review, with extra time for the assumed exception rate.
• Other effort: recurring supervision, configuration upkeep and support; show one-off migration separately.

03Read the scenario as arithmetic, not a product result

Suppose 100 documents currently take five minutes each, including review: 500 minutes. In a hypothetical pilot, assume one minute of preparation and two minutes of routine review for each document. That is 300 minutes. If 20 documents each need another four minutes of exception work, the scenario total is 380 minutes. The difference is 120 minutes, or two hours, before other recurring costs.

Those figures are illustrative assumptions, not measured Vacei performance. Replace them with your observations. A negative difference is useful too: it shows that the chosen workflow, configuration or client cohort may need work before expansion. Do not turn estimated hours into revenue without separately considering demand, pricing and available reviewer capacity.

04Run a pilot with clear acceptance criteria

Choose a small cohort that reflects the files you expect to process, including a few messy but manageable cases. Agree the start and end points of the workflow, who reviews it and which outcomes matter. Compare like-for-like periods and keep a record of changed assumptions. A month with fewer transactions or cleaner source records is not automatically evidence of software improvement.

Before expanding, discuss both the time record and the quality of the completed work. Were unsupported entries left visible? Could the reviewer reach the source document? How much rework remained after review? Your acceptance criteria should describe observable behaviour rather than a broad promise to save time.

05Bring your scenario to the software demonstration

Vacei's published firm overview describes connected accounting, audit, practice and client portals, with AI preparing routine work and people reviewing and signing. Its business pages label their timing examples as illustrative rather than measured results. Use the same distinction when assessing your own firm: product descriptions show what to investigate; your pilot establishes what happens on your files.

Bring aggregate volumes, your workflow diagram and the assumptions you want to test to a firm demo. Agree an appropriate sample-data process before sharing client records. Ask for the review path, exception path and export path to be demonstrated. That gives your team a practical basis for deciding which workflow to introduce first.