Policy, compliance and governance support for the care industry
CareScope publishes your policies under a named review, answers your staff from them with the exact source lines, and turns what the policy doesn't say into owned, published and independently verified work.
Support for the policy and governance work your team already does. It does not certify care quality or compliance, and it does not score you.
Where it shows up first
2am questions. We got you.
Night shift. New starter. Agency cover. The office is shut, but the questions aren't.
At 2am, staff ask the Care Desk. It answers from this home's published policy and shows the lines it came from. If the policy doesn't say, we don't guess.
For the registered manager who signs the policy off, and everyone working the shift.
One question, all the way through.
Ask › Source › Gap › Owned › Published › Verified. Open a step to see the surface it happens on.
Care Desk / Find your local policy
Your policy question
What is the leave request procedure?
Ask about the policy, without personal details. Each question stands alone.
Ask question
Validated answer
Use the leave request form.
Assistant explanation, not the original policy wording. Check the sources below.
Staff sign in to their own Care Desk and ask in their own words. The assistant may only use this home's currently published release. Not a general model answer, not another home's wording, and not a draft nobody has approved.
Answer text is withheld until source validation completes, so a half-finished procedure is never shown as guidance.
Nothing on this page submits a question, creates an account or produces a live answer.
Recorded sources
Local procedure: Leave requests
Use the leave request form.
Policy [policy ID] · Release [release ID] · Lines 5-5 · Commit [commit SHA]
Quotations are your home's published wording, delivered to the assistant as stored.
Saved receipt [receipt ID] / Your question and evidence
Every citation is a structured policy ID, path, release, commit, heading, line span and exact quote, checked against the release snapshot the question was answered from. A search snippet, a plausible filename or a confident sentence is not a source.
The receipt keeps the question, the answer, the source lines and the exact policy commit. Publishing a newer policy tomorrow does not rewrite what was answered tonight.
A source check proves provenance. It does not prove an explanation is clinically correct or right for every situation.
Staff question
“Where do I send the leave form?”
Refused / not found in approved policies
The test policy says to use the form. It does not say where to send it.
Request policy clarification
I confirm this is a minimised policy clarification, not an urgent incident report.
A missing instruction is recorded as missing. Staff raise it from their own saved receipt, so the office gets the question that was actually asked instead of a retelling in the corridor, and they choose the minimised concern to share.
A miss is not automatically a confirmed policy gap, and a technical failure is not a training need. Urgent incidents and safeguarding still go through your existing routes.
This is not a confidential safeguarding channel or an urgent incident reporting route.
Triage and assign · Assigned
Give this concern a named owner, a due date and the criteria that will count as done.
Assign this work
Audit timeline
[date] / [actor name] / concern recorded
[date] / [actor name] / assigned
Append-only events, not an editable note field.
A gap on the floor lands in the office with a state, a named owner inside your home's scope, a deadline and an agreed definition of done. Your daily agenda shows what is unassigned, what is overdue and what is waiting to be verified.
There is no decorative score on top of it. You open the actual work behind each agenda line.
Bracketed values are placeholders. No real person, home or deadline is shown.
Unpublished draft · not available to staff or staff search
Named review and explicit decision
Compare the preserved original with its converted preview. If conversion omitted instructions, correct the source and import a new draft before approval.
Approve and publish reviewed wording
Record completed work
Completion needs reviewed evidence and the published policy section that now covers the gap.
Evidence [reviewed evidence] · Release [release ID] · Section “## Leave requests”
An imported document is an unpublished draft until a named manager reviews the converted wording and publishes it with an effective date and a review due date. A replacement supersedes the previous instruction and keeps its lineage.
Releases are Git-backed, so the fix has a commit, and the completed work points at the published section rather than at a reassuring message.
A recorded review is not proof of clinical correctness.
Verification decision · Effectiveness verified
A different scoped manager checks the completed work against the acceptance criteria.
Decision: Verify effectiveness · Return for more work · Do not pursue (reason required)
Observation: [recorded observation]
My policy reading inbox
Reviewed change note + the exact published version
Unread versionReading recorded for this version
The requester can read the published clarification and acknowledge the update in the app.
The owner or the person who completed the work cannot verify it. Another authorised manager checks it against the acceptance criteria and can return it for more work.
Then the change lands where the shift can see it: the reading inbox pins the exact version, and a manager-reviewed procedure card carries the Do / Don't / Tell / Record line straight from the approved source span.
A reading acknowledgement is not evidence of competence.
Your policy. The exact lines. A receipt to come back to.
What goes to the model? What stays stored?The bit a policy folder can't do
A gap on the floor becomes owned work in the office. Not another message for you to remember.
No answer in the policy? Nothing is invented. Staff can request policy clarification from their own receipt and choose the minimised concern to share.
Where should the leave form go?
Management update, not approved instructions. The approved wording is the published policy, and the register links to the exact section and commit.
Illustrative route to a fix / each step leaves a record
01 / Awaiting triage
Concern: the leave form has no destination. A manager checks what is missing.
02 / Assigned
03 / Work completed; verification pending
A manager publishes the missing instruction. Link the policy section and completion evidence; put the reviewed change in the reading inbox.
04 / Effectiveness verified
Another authorised manager checks the fix against the acceptance criteria. If it doesn't work, return it for more work.
Completed and verified are separate states. The owner or person who completed the work cannot verify it. The requester can read the published clarification and acknowledge it in the app.
A miss is not automatically a confirmed policy gap. A technical failure is not a training need. Urgent incidents and safeguarding still go through your existing routes.
On the floor / in the office
Decide what goes live. See what needs attention. Keep the work moving until someone checks the fix.
Open a surface
Preserved original
Compare the file with its converted preview. Check for omissions.
Named review and explicit decision
Reviewer · change summary · effective date · review due date
Approve and publish reviewed wording
Release history and version comparison stay available afterwards. Withdrawal removes an instruction from current authority.
Improvement register
Concern: where should the leave form go?
Owner: [named manager]
Due: [deadline]
Staff-confirmed, minimised concerns. Triage, assign, complete, verify. Every transition writes an append-only audit event with the actor and the source commit.
This register is not urgent incident reporting.
Questions & clarification work
Record · Submitted · Outcome · Completed
Review the question stream by outcome: validated answer, not found in approved policies, source validation failed, service failure, processing not completed. Manager preflight and test-provider runs are labelled and excluded from staff activation.
Analytics does not display original questions or answers.
See which staff have recorded reading against which exact published version, and who asked for clarification on it. Acknowledgements bind the policy identity, content hash, release and commit, so an earlier acknowledgement is not overwritten by a later publication.
A reading acknowledgement is not evidence of competence.
Invitations are single-use and expiring. The recipient chooses their own password, so a manager never holds a usable staff credential.
Individual accounts. Explicit home access. No home grants means no home access.
Asked, answered, missing and fixed stay separate, and an approved governance review freezes the evidence versions, exact policy commits, interpretation and outstanding action IDs you actually reviewed.
See the review snapshotYour next decisions
Open the work behind each agenda item. No decorative score.
Recorded work for this home, not a compliance score. Completed work stays separate from verified effectiveness.
Back on the floor
Web-based, with nothing to install. The current wording is there to read, even when staff aren't asking the assistant.
Reviewed change note + exact published version
Managers see Reading & clarification records. A reading acknowledgement is not evidence of competence.
Do: “Use the leave request form.”
Local procedure › Leave requests · Line 5
Don't / Tell / Record: Not specified in this card.
Manager-reviewed source excerpts. A changed source disables the card until a new review. Read the full policy and exceptions.
Take it into your review
What staff asked. What came back. What was missing. What you fixed.
Separate records, not a reassuring percentage. Follow the work from the question to the verification.
Recorded staff interactions. Receipts retain the original question for authorised review.
Validated answers, with source lines tied to the published release.
Staff-confirmed concerns and triaged gaps, not every unsuccessful request.
Published clarifications, completion evidence and separate effectiveness verification.
Freeze what you reviewed.
Later edits do not rewrite this approved snapshot.
Report PDFs are hashed and checked again on download. Missing, tampered or wrong-home files fail the check. If rendering fails, the report is marked failed. No HTML stand-in.
There is no overall compliance score, no competence score and no readiness percentage. Reports support your review; they do not certify care quality or compliance.
Validated, refused, ungrounded, technical error and pending are recorded separately. Refused means not found in approved policies; ungrounded means source validation failed. Technical errors are service failures; pending is not a completed outcome.
Mock runs, management questions and manager preflight are not staff activation and are excluded from production staff outcomes. A source check proves provenance, not that an assistant explanation is clinically correct or suitable for every situation.
CQC publication monitoring concerns England and does not establish comprehensive UK-wide regulatory coverage. Not assessed, unavailable, stale or nothing new recorded are not a pass.
A future review date is not a completed review. Reading a policy is not evidence of competence. Policy coverage and practice evidence remain separate.
Straight about privacy
Names, dates of birth, contact details, NHS numbers and similar identifiers are stripped from the question. That detected personal detail does not go to the model.
Detected name → [NAME] · Detected DOB → [DATE OF BIRTH]
Automated detection can miss an identifier or an indirect clue. It reduces risk rather than proving a narrative is anonymous.
Published policy is not put through that filter. Names, addresses, dates and schedules written in it reach the model intact. Those are your published instructions.
Do not publish resident records as policy.
The original question is still stored for authorised review. This is not a confidential safeguarding channel or an urgent incident reporting route. Do not submit resident information to evaluate the service.
Processing location, retention and processor terms still need agreeing for your deployment.
Read data handling and outstanding privacy requirementsDiscuss a pilot
Let's work out a pilot around your home, your staff and the instructions they need to find.
Pricing, availability, trial duration and service commitments must be agreed separately. Registering interest does not start a trial, create an account or purchase an entitlement.