There are 2 kinds of accounting automation, and only 1 of them ever produces a large number.
The first shortens a task. Faster capture on an invoice, a better rules engine suggesting a coding, a tidier approval inbox. A person still performs the work with fewer keystrokes, and the saving has a floor: however long it takes somebody to look at a thing and decide it is fine.
The second deletes the task. Nobody uploads, nobody keys, nobody reviews, because the step is not there any more. Almost every figure worth quoting in this category comes from the second kind, and almost every product demo shows the first.
less time on the books at All Gravy's bookkeeping partner, while the company doubled
card transactions a month at All Gravy that used to be uploaded by hand
average time saved on repetitive admin work at teams running a Custom Agent
Two kinds of automation
The first kind shortens a task. Faster OCR on an invoice, a better rules engine suggesting a coding, a cleaner approval inbox. The person still performs the task, with fewer keystrokes. Savings are real and they are bounded, because there is a floor: the time it takes a human to look at a thing and decide it is fine.
The second kind deletes the task. At All Gravy, 500 to 600 card transactions a month were uploaded by hand, and a single missing receipt held up an entry while somebody tracked exceptions across 2 systems. That number is now zero. Not faster uploading. No uploading, and no exception-tracking process either, because the exceptions were a property of the manual pipeline rather than of the transactions.
Almost every large number in finance automation comes from the second kind. The first kind produces 20%. The second produces the figures on this page.
Where the deletions are
They cluster in the same 4 places at every company, which is what makes them worth naming.
Bills arriving. At All Gravy bills come in by email and Light's AI reads and codes them on arrival. Nobody opens them, nobody keys them in. At Ocean.io a bill is split into the correct P&L accounts by rule before anyone looks at it.
Invoices going out. Tillo pushes 750 to 900 invoices a month straight over the API with no review step, and revenue lands in the books on T+1. At Famly, a 20 hour revenue process now takes 15 minutes.
Money arriving. Reconciliation runs continuously against live feeds under the Bank Rec Agent's own identity, rather than as a monthly matching exercise.
Chasing. Collections runs as an agent that tracks outstanding balances across every entity, reads the correspondence history with each customer, and drafts and sends the reminders.
"The dunning agent gives us the full picture of what is outstanding and handles the chasing. That used to be my afternoon."
Linea Meldgaard Andersen, Finance Analyst, All Gravy
Shortening
The task gets faster
Better capture, better suggestions, a tidier queue. A person still performs the work, and the floor is how fast a human can review.
Deleting
The task stops existing
Card spend lands in the ledger untouched. Contracts issue their own invoices. Nothing accumulates into a queue for anyone to work through.
Writing
The team automates the rest
Work specific to 1 company gets described in plain language and handed to a Custom Agent. Average time saved on that work: 80%.
The ceiling is higher than most teams assume
At Dreamdata the figure is 95%.
"We've automated 95% of bookkeeping with Light and some simple custom agents. This thing is powerful."
Andreas Jørgensen, Controller, Dreamdata
The relevant detail is the second half of that sentence. Some of the automation came with the platform and some of it Andreas wrote himself, in plain language, for jobs specific to how Dreamdata's books work. Across teams running these, the average time saved on that category of repetitive admin work is 80%.
That is the pattern worth copying. The shipped agents cover the work every finance team shares. The last stretch, the part that gets a company from 80% to 95%, is written by the people who know that company's chart of accounts, because nobody else can.
What the freed time is for
The point of removing the work is what fills the space.
"We want to remove the time spent on pure transactional work and move it to strategic planning, budgeting, and forecasting. Getting accuracy in the forecast means we can be certain we are spending where we should be."
Sebastian Sandorff Jacobsen, Head of Finance and Ops, All Gravy
Peter Egehoved, COO and CFO at Dreamdata, puts the same thing more briefly: the agents "allow me time back to do more financial planning and strategy." At Tillo, Harriet Stewart says finance "stopped being a boring back-office function and became a key player in driving growth and direction."
That is 3 companies describing the same shift in 3 ways. A finance function stops operating a process and starts running an analysis, and the people in it stop being the constraint on how fast the business can grow. All Gravy doubled with a team of 3. That is the number the automation actually bought.
See the agents that ship with Light, the Custom Agents any admin can write, or book a demo.