Work it out yourself

Your volume. Your cost. Your number.

This is the same arithmetic we do in a scoping session, run in your browser. Nothing is sent anywhere, there is no form, and we do not know you were here. If the number is not big enough to be worth a meeting, you have saved yourself the meeting.

No form · No email · Nothing transmitted

Your numbers

Defaults are the midpoint of what we see in each vertical. Overwrite them with yours — the result updates as you type.

Sets the default handling time and complexity rate. It does not change the maths.
Whole files, not pages — a loan file, a claim, an inspection pack.
End to end: retrieving, reading, checking, keying and writing it up.
Salary, benefits, tax, supervision and floor space. Not base salary.
Densery's third question: what happens when the output is wrong. This is where the number usually surprises people.
Re-review, downstream correction, and the expected cost of the tail — a reserve development, an exam finding, a customer credit.
We deliberately default below what our deployments run at. Move it down if you want the pessimistic case — the conclusion rarely changes.
On-premise adds a fixed annual platform fee for the appliance and its support.

Projected impact

Year one, at your numbers

net annual benefit after what you pay us

What the work costs you today
labour on review, plus the cost of the files that come back
Reviewer hours returned
What you would pay Densery
per completed file, not per page
Return on every dollar spent

Runs entirely in your browser. No values leave this page — check your network tab.

Every assumption behind that number, and where we think it is weakest

A calculator that hides its arithmetic is a brochure with a slider on it. Here is all of it.

AssumptionValue usedWhy, and how it could be wrong
Productive hours per reviewer per year1,6002,080 gross less leave, training, meetings and administration. If your team is heavily part-time this overstates the hourly cost and therefore the saving.
Supervision time on a completed file8% of the original handling timeSomeone still reviews exceptions and samples the rest. In the first quarter of a deployment this is realistically 15–20% while thresholds are tuned, so year one is worse than steady state.
Files not completed by agentsCost unchangedThey still take the full manual time. We do not claim partial credit on files the agent could not finish — that is the assumption most vendor calculators quietly break.
Rework reduction60% of your rework costThe evidence layer catches its own errors before they leave the process; it does not catch errors that were in the source document. This is the softest number on the page.
Price per completed fileIllustrative planning rate by complexityDemo placeholder. Real pricing depends on document type, integration surface and volume commitment. Replace this with the live rate card before the page ships.
On-premise platform feeUS$180,000 per year, illustrativeAppliance, support and model updates. Also a demo placeholder.
Implementation costExcludedDeliberately. Payback below is on run-rate only; a first-year view should carry the implementation fee against it.

This calculator is a planning tool, not a quote, and nothing on this page is a contractual commitment. If the number matters to a decision, bring twenty real files to a scoping session and we will rebuild it against measured handling times rather than your estimate of them.

What the number does not tell you

The saving is the easy part of the business case.

The evidence burden is the blocker

88% of agent pilots never reach production. Almost none of them fail on arithmetic — they fail because risk and audit cannot see how an individual decision was reached. That cost does not appear above, and it is the one that kills projects.

Cycle time stops being a hiring question

The number above is annual. The change your operations lead will actually feel is that a backlog becomes a configuration change rather than a recruitment cycle — which matters most exactly when volume spikes.

The unit cost keeps falling

Every correction a reviewer makes feeds back into your ontology, so the automated share rises on your own documents over time. The calculator holds it flat for year one, which makes it conservative by design.

IF YOU WANT THE HARDER PROOF

A number you calculated yourself is still a number about money. The question your risk committee will actually ask is how a decision was reached. Open a real file in the audit-trail explorer →

The next step

Ninety minutes, your documents, three numbers.

A scoping session is not a demo. Bring twenty real files, redacted if you need to. We take three numbers off you — annual volume, fully loaded cost per file today, and what happens when the output is wrong — and hand back a one-page value case in your own KPIs.

If the arithmetic says we are not a fit, we will tell you in the room rather than six weeks later.

QUALIFY YOURSELF OUT

We are a fit if all three are true

  • More than 250,000 pages a year, or 25,000 claims or files
  • A legal obligation — regulator, board risk committee or parent-company policy — to keep the data in-house
  • An AI or agent pilot that did not reach production

If your data can go anywhere and your documents are already clean and digital, you do not need us. Use a hyperscaler document API and spend the money on something harder.