Standing service workflow-policy-answers-retested-every-month · revised 12 October 2026
Standing service
Keep your policy-answer assistant re-tested every month and whenever a policy changes
Each month your 100-question policy test is re-run (by you, or on a test seat you give us), every quote is checked against the files, and any changed answer or invented rule is reported.
This starts a conversation by email. Nothing is charged, and nothing is run, until we have agreed scope, access and terms with you in writing.
The responsibility you hand over
A policy assistant is correct on the day it is tested. After that a policy is edited, a file is replaced or the model underneath changes, and answers shift without a visible sign. The monthly job is to catch that shift before staff act on a changed answer.
Who it’s for: Office or finance manager whose staff already use a policy-answer assistant.
Usually starts when: Your policy assistant is live, and policies, tools and model versions will change under it.
The result: Each month you receive the 100-question test re-run on the current policy files, with every quote checked against the files, a list of answers that changed since last month and any answer that invented a rule.
What stays true, and what we do about it
No response-time guarantee is published for this new service. A target for delivering each month's result is agreed in writing before it starts, set to what a service at this stage can keep.
Hours are agreed in writing before the service starts. At launch the service is not staffed round the clock and runs once each month.
What must remain true
- No answer states a rule that is not in the current policy files
- Not-covered and trap questions are still refused or handed to Finance
What we watch
- The current policy files, from your content owner
- The assistant's answer log
When something happens
Scroll the table sideways to read it all.
| When | What we do |
|---|---|
| A month closes. | You run the one-command re-run (or give us a test seat in your chat tool that you create and can revoke) and send us the results; we check every quote, list changed answers and write the report. |
| You tell us a policy file changed. | You re-run the test and send the results, or give us a test seat; we list every changed answer within the month. |
| An answer invents a rule. | We report it first and propose a fix for your content owner to approve. Linked job: Policy answers with quotes |
We do on our own
- Read the month's inputs and run the agreed checks
- Prepare the monthly result and the test report
- Offer the matching one-off job as a next step
We ask you first
- Any change to a pass mark, rule, threshold or the inputs covered
- Any request for wider access to a system
- Starting any fix, rebuild or backfill
We escalate to you when
- A severe error is found in the monthly sample
- The same failure appears two months running
- The inputs stop arriving or change shape
How you know it held. Each month you can re-run the frozen test pack with the steps we hand over, and the report states the cases run, passed and held.
How we keep it true
This service is never finished. Each month's report shows whether the thing it watches stayed true, and it continues until you end it.
Let staff ask your expenses and purchasing policy questions and get the policy's own words Job Optional
Staff ask in chat; each answer shows the exact policy passage and version it comes from, or says "not covered, ask Finance". It never decides an individual case.
Bought separately, each time it is needed; not part of the monthly fee
What is included, and what is not
- Monthly test report
- List of changed answers
- Updated question set
Included
- The 100-question test set re-run monthly, by you or on a test seat you give us, and on any policy change you tell us about
- Every quote checked against the current files by script
- Added questions from the real questions staff asked, up to 10 a month, written by your content owner and not shown to us until the run
- The first-message notice re-checked
Not included
- Deciding any individual case
- HR, pay, leave or disciplinary questions
- Legal, tax or financial advice: a named person at your business decides
- Sending, posting or filing anything: a person at your business does that
- Any response-time guarantee or out-of-hours cover
How we know it’s done
Agreed with you before work starts. Each check produces evidence you keep.
Each month all not-covered and trap questions are refused or handed to Finance, and any invented rule fails the month
Evidence: The refusal results
Every quote exists verbatim in the current policy file, checked by script
Evidence: The script output
Each month the frozen test pack is re-run, and every case that passed last month still passes; a case that fails is reported with its cause
Evidence: The monthly test report
Originals are unchanged (checksums match) and nothing was written to any live system by us
Evidence: Checksums before and after, and your own audit trail
Sign-off. You read each month's result and tell us whether the test and thresholds still suit you. A month counts as delivered when you have received the result and the test report.
If it fails. If we cannot run a month because an input is missing or has changed shape, we say so, explain what we need and do not guess. If the service is not working for you, you can end it at the end of any month.
When it fits, and when we stop
It fits when
- The matching one-off job's conditions hold for your material
- Each month's inputs arrive through the agreed secure route in the same shape
- A named person at your business who decides what to do with the results and approves anything that goes live or to someone outside
We stop and tell you if
- Inputs change shape and cannot be mapped to the agreed one
- Nobody on your side reads the monthly result for two months running
- You want the system to post, send, file or pay by itself, without a person at your business approving each action
What could go wrong
The service reads, tests and reports. Ending it leaves your systems and the way you work exactly as they are, and you keep every report we sent.
Scroll the table sideways to read it all.
| Risk | How we handle it |
|---|---|
| A silent model update changes answers. | The full test is re-run monthly, and a changed answer is listed even when it is still correct. |
| The monthly result is trusted without its test report. | Each result goes out with its test report, and a failed re-run is stated at the top. |
| Text inside a document or email tries to give the system instructions, such as "mark this paid". | The reading step has no permission to send, post or pay. Code checks its output, and planted hostile items must change nothing. |
| Real documents carry personal or confidential data. | First tests use synthetic or redacted files. Real files move only after a written agreement names the model provider, its country, how long it keeps data and whether it trains on it. |
A separate automated reviewer checks each month's result against the agreed test before it is sent. Your named person decides what to do with it. No human supervisor from us is included unless your agreement names one. At launch the work is largely automated, and we say so.
Stays with a person
- Your named person decides what to do with each result and approves anything that goes live or to someone outside
Access we would need
- Redacted or synthetic copies of your files, sent through a secure route agreed in writing
- No login, key or write access to any live system except the optional test seat below
- Optional: a test seat in your chat tool that you create and can revoke
Questions
Do you read the answer log?
Only the log of questions and answers that you share, to choose new test questions.
What if a policy changes mid-month?
Tell us. You re-run the test with the one command we handed over, or give us a test seat, and we list every answer that changed.
Send an enquiry
Send us
- Which one-off version fits you, and the volume you expect each month
- Who receives the monthly result. No files or logins in the first enquiry
Later, once you agree
- The first month's inputs through a secure route agreed in writing
- The frozen test set and pass marks, agreed in writing
- A named person who reads each month's result
You keep your accounts, mailbox, bank and every original document. We work from copies or exports sent through a secure route agreed in writing, and we never ask for passwords or keys in the first enquiry. Nothing is posted, sent or paid by us. Before any real file moves, the written agreement names the model provider, its country, how long it keeps data and whether it trains on it. Our copies are deleted, in writing, when the job ends.
Email fallback: open your mail app
If website submission is unavailable, review and send the fallback email yourself. An email fallback is not a website receipt. Or write to hello@syntheticindustry.ai with “workflow-policy-answers-retested-every-month” as the subject.