What we do
Results as a Service: you pay for each result that passes its checks.
We build an AI agent for one job your team does by hand. After launch you pay a fixed price for each result that passes the checks you signed. Work that fails, or that your team finishes, is not billed.
Work out your own billWhat it costs
$10,000to build, paid on acceptance
After launch, a fixed price for each result that passes, set from your acceptance test.
- Each result that passes
- fixed price
- Monthly minimum
- $1,200
- Results that fail a check
- $0
Who we did it for
- Rockwell Automation: Run on results-based terms. Finding information went from 12 minutes to under 4.
- Generac: An assessment run on the same terms, with our fee tied to the results we agreed.
- Pharma and life sciences: Several clients in the industry pay us this way. Their names stay confidential.
Most of our projects are paid against results agreed up front. 200+ production systems shipped since 2014.
Case studiesWhat we want from you
Tell us one job you would pay for by the result.
Results as a Service (RaaS) is how we charge for an AI agent once it is live. You pay a fixed price for each finished result that passes the checks you signed, such as a matched invoice or a drafted certificate. Anything that fails a check, or that a person on your team has to finish, is not billed. You pay the $10,000 build fee only after the agent passes your acceptance test.
Why the usual ways of paying for AI leave the risk with you.
Seats you pay for either way
A licence costs the same in a month when nobody opens the tool, and the same again when it gets answers wrong.
Hours that grow when work slips
With hourly billing, a build that runs late costs you more, and the firm that ran late is paid more for it.
Charges for attempts
Per-conversation and per-action pricing bills you each time the AI tries, including the tries that fail and land on your team anyway.
A flat fee after launch
A flat support fee is the same whether the agent finishes 300 results or 30, so nobody is paid to raise the pass rate.
From one workflow to a monthly bill you can check line by line.
- 01
Define one result
Scoping callWe agree what one finished result is, for example a supplier invoice matched to its purchase order and receipt with any variance explained. We write down the checks it must pass and who on your team approves it.
- 02
Build and run the acceptance test
Weeks 1 and 2We build the agent in your environment and run it on a sample of your real work. The $10,000 build fee is due only if it passes the acceptance criteria you signed.
- 03
Fix the price per result
Day 14The test gives us the pass rate, the volume and the time a person still spends on each result. From those numbers we agree one price per result and write it into the contract with the $1,200 monthly minimum.
- 04
Run, count and bill
Every monthEach result is logged with the checks it passed and the person who approved it. Your bill is the number of passing results times the price, or $1,200 if that is higher. You can open any line and see why it was billed.
- 05
Reprice from real data
Once a quarter, if neededIf volume or the pass rate moves a long way, either side can ask to reset the price using the last quarter of logged results.
What is one result worth to you?
Put in your own numbers. The calculator shows what the work costs you now, what a month on Results as a Service would cost including the work that stays with your team, and the highest price per result that still saves you money. Model usage and hosting run on your own cloud bill and are not included.
You would save each month
$4,781
149 staff hours back a month. The $10,000 build pays for itself in 2.1 months.
- The work by hand today
- $9,000
- Our bill: 1,275 passing results at $1.50
- $1,913
- 225 items that go to your team, not billed
- $1,350
- Your team reviewing passing results
- $956
At these numbers, any price per result under $5.25 saves you money.
Get a real price per resultWhat counts as a billable result.
The contract lists the checks for your workflow. These rules apply to every one of them.
| What happened | Why | Billed |
|---|---|---|
| Passed every check and your reviewer approved it | This is the finished work you are paying for. | Yes |
| Failed a check | The agent sends it to your team with the reason it failed. | No |
| Outside the agreed scope | The agent hands cases it was not built for to a person and does not count them. | No |
| Your reviewer rejected or rewrote it | If a person has to redo the work, it was not a result. | No |
| A duplicate or a test run | Duplicates and test items are filtered out before counting. | No |
We have worked on results for years. Now it is how we charge after launch.
Most of our projects are paid against results we agree with the client before we start. Our work for Rockwell Automation ran on those terms, and so did an assessment for Generac. Several pharma and life sciences clients, whose names stay confidential, pay us the same way.
Results as a Service carries that into the months after launch. A flat support fee pays us the same whether the agent finishes 300 results or 3,000. A price per result pays us more only when the agent does more of your work correctly, so raising the pass rate matters to us as much as it does to you.
There is a trade-off. Your monthly bill moves with your volume, and finance teams usually prefer a fixed number. The fixed price per result and the monthly minimum keep the bill inside a range you can forecast, and every billed line can be checked against the log.
Jobs where one result is easy to count.
Results as a Service works when a finished result is easy to recognise and check. These are typical first workflows.
Supplier invoice matching
One result: an invoice matched to its purchase order and receipt, with any variance explained.
Certificates of insurance
One result: a certificate drafted from the policy and the request, ready for your CSR to send.
Open →Promotional material review
One result: an asset checked against your claims rules with every flag cited, ready for your review committee.
Open →Support requests
One result: a request answered from your knowledge base and closed without a person stepping in.
Prior authorization packets
One result: a packet assembled with the clinical evidence the payer asks for, ready for a nurse to sign off.
Results as a Service next to the other ways to pay for AI.
| How you pay | What the fee is tied to | If the AI underperforms |
|---|---|---|
| Per seat or licence | Everyone who has access, used or not | You pay the full fee |
| Hourly consulting | Time spent building and fixing | You pay for the extra hours |
| Per conversation or action | Each time the AI runs | Failed attempts are billed too |
| Flat monthly support | Upkeep, whatever the output | You pay the same amount |
| Results as a Service | Each result that passes its checks | Failed results cost nothing |
What the monthly bill covers.
A log of every result
Each result is stored with the checks it passed, who approved it and how long it took. Your bill is built from this log.
Monitoring and fixes
We watch the pass rate and fix what breaks when a form, a supplier or a rule changes.
Model and prompt updates
When a better model arrives, we test it against your acceptance checks before we switch.
A monthly report
Results passed, results sent to your team, hours returned and the bill, on one page.
Your data stays with you
The agent runs in your cloud tenant. We build systems to SOC 2, HIPAA and GDPR standards.
From workflow to working tool in two weeks.
Scope and sign
We define the workflow, the deliverables, and the acceptance criteria, and sign an agreement on them before anything starts.
Build
We build the tool in your environment, with Claude Code and automated evaluation against your own data.
Pay when it works
$10,000, due only after every criterion in the signed agreement is met. Nothing before that.
The build is paid only after it passes your acceptance test. After launch you pay for results that pass their checks, with a $1,200 monthly minimum. Work that fails a check or that your team finishes is not billed.
Tell us the workflow →Questions buyers ask.
What is Results as a Service?
Results as a Service is a way of paying for AI where you pay a fixed price for each finished result that passes agreed checks, such as a matched invoice or a drafted certificate, instead of paying for seats, hours or a flat monthly fee. At QueryNow it applies after launch. The build is $10,000, paid on acceptance.
How is the price per result set?
From your acceptance test. We run the agent on a sample of your real work and measure the pass rate, the volume and the time a person still spends on each result. The price is written into the contract before the first bill.
Why is there a monthly minimum?
The $1,200 minimum covers monitoring, fixes and model updates however many results there are in a quiet month. If your passing results add up to more than $1,200, you pay for the results and the minimum is not added on top.
Who decides whether a result passed?
The checks you signed decide, together with your reviewer's approval. Both are logged against every result, so each billed line can be traced back to what happened.
Do people on my team still review the work?
Yes. A person on your team approves each result before it goes to a customer, a regulator or your ledger. The calculator on this page includes that review time so the saving is not overstated.
Does the price per result include model and cloud costs?
No. The agent runs in your own cloud tenant, so model usage and hosting appear on your existing cloud bill. We estimate both from your real volume during scoping, before you sign anything.
Which jobs are a poor fit?
Work where a good result is a matter of taste, or where volume is a handful of items a month. If you cannot write down how to check a result, it is hard to price it per result, and a fixed build is usually the better choice.
Do I still pay the $10,000 build fee?
Yes, but only after the build passes the acceptance criteria you signed. If it fails, you owe nothing for the build and there is nothing to run.
Ready to ship AI that actually works?
Send us the workflow. We return a fixed scope, price, and acceptance criteria in 48 hours. You pay $10,000 only after it works.
Tell us the workflow →Buying AI by the result: a Results as a Service guide for operators and finance teams
How to define, check and price AI results, with published per-result prices and the contract terms to ask for.
Read it free →