The last week of every month, given back.
Most mid-sized businesses rebuild the same reports by hand every month: export, paste, reconcile, explain, send. An agent does the assembling and the first draft of the explaining, and your people do the judging.
Reporting by hand
- Figures exported from four systems and pasted into one spreadsheet
- Numbers that don't match last month, with no one sure why
- Commentary written at 10pm the night before the meeting
- Reports that arrive after the decisions have already been made
Reporting by agent
- Figures pulled straight from the source systems, every number traceable
- Variances found and explained before anyone asks
- Commentary drafted in plain English for the owner to check and edit
- Packs delivered on schedule, with a record of who approved what
How a report builds itself
- 1 Pull Figures come straight from finance, CRM, operations and support, on a schedule.
- 2 Reconcile Every number is checked against its source and against last period.
- 3 Explain AI drafts the commentary: what changed, why, and what deserves attention.
- 4 Approve The report owner reads, edits and approves before anything is sent.
- 5 Deliver Packs go out by email, portal or board platform, with an audit trail.
Watch a report build.
An illustrative run of the kind of reporting agent we put into production. The figures are invented; the way it works is not.
The report that arrives.
A page from the kind of management report this produces. The numbered markers show how each part got there.
- Pulled from the source systemsEach figure carries a link back to the record it came from, so anyone can check it.
- Explained by the agentThe commentary is a first draft, written from the transactions and notes behind each movement.
- Approved by a personNothing is sent until the document's owner has read it, edited it and signed it off.
Monthly management report
Revenue by month, $m
What changed and why
- Revenue rose 6.2% on September. Most of it is the 1 October price rise landing in full.
- Margin fell 1.2 points because a supplier invoice dated September was booked in October. Without it, margin is flat.
- Two renewals worth $94k a year fall due in November. Both account owners have been told.
Illustrative. Names and figures are invented.
Watch a family office get one view of everything.
A 72-second film, with no sound to turn on: hotels, property, private funds and shares reported from eight systems, rebuilt by hand each quarter, then collected overnight into one reporting store and one position.
Read what the film shows
- By hand, each quarter. A family office holds hotels, investment property, private funds, listed shares and cash across six entities. Each asset reports in its own system and format: ledgers, a custodian file, fund portals, bank portals, hotel systems, property managers' statements and payroll. Every quarter someone rebuilds the group position in a spreadsheet. It takes three weeks, the figures are as at five different dates, and one person knows how it all fits together.
- Overnight, with agents. Agents collect from every system each night, check each figure against its source and fit it to one model of entities, assets and periods. What ties out is loaded into one reporting store that the agents maintain. What does not, such as a difference between the custodian and the ledger, goes to a person with the documents attached.
- One view. A consolidated position as at that morning: net assets, cash and the split across property, hotels, private funds, listed shares and cash. Hotels show occupancy, rate and labour cost side by side. Property shows rent, leases and valuations. Private funds show what is committed, called and still to come. A twelve-month cash forecast flags the month cash falls below the family's floor. The principal's question is answered the morning it is asked, and the committee pack is drafted for the CFO to review.
Where the AI comes in.
Pulling numbers into a template is ordinary automation. The parts that used to need a person are where AI earns its place.
Explaining the numbers
The agent writes the first draft of the commentary: what moved, by how much, and the most likely reasons, drawn from the underlying transactions and notes.
Spotting what's wrong
Figures that break a pattern, duplicate entries and totals that don't tie out are flagged with a plain-English note, not buried in a cell.
Answering the follow-up
When a director asks why travel costs rose, the answer comes from the data behind the report, in seconds, with the records to back it.
Built to run in production, like everything we make
- Inside your systems Agents work in the CRM, finance and operations tools you already run, with the same permissions as a member of staff.
- A person approves what matters Payments, customer messages and anything above a limit you set wait for a human yes.
- Every action logged Each step is recorded, so you can answer what happened, when and why.
- Measured and handed over Results are measured against today's baseline, and your team is trained to run and change it.
Questions people ask us
What is automated reporting with AI?
Automated reporting with AI pulls figures straight from source systems such as finance, CRM and operations, reconciles them, writes plain-English commentary on what changed and why, and delivers management, board or client reports on schedule after a person approves them.
Can AI write the commentary in a board pack?
Yes, as a first draft. The agent explains movements from the underlying transactions and notes, and the report owner checks and edits it before anything is sent. Every figure stays traceable to its source.
Which reports are best to automate first?
Reports that are rebuilt by hand every month from several systems: management reports, board packs and client review reports. A discovery session is the quickest way to find the one that will free up the most time.