The care record that answers questions.
Your home already writes everything down. Verity is the first UK care platform that can read it back — ask in plain English across a resident's whole history and every clause of the answer is footnoted to the note, timestamp and author behind it.
A 48-bed home writes care records a year. Almost none of them are ever read again.
UK care homes record 50 to 65 notes per resident per day. When something goes wrong, the reason is usually already written down — by four different people who each saw one piece of it, on four different days, in four different words.
The sector digitised its records and did not gain the ability to read them. These are findings from the independent NIHR-funded evaluation of digital social care record rollout in England — not our marketing.
Inability to retrieve data from more than a week ago.NIHR rapid evaluation of DSCR implementation in England
Navigation to necessary forms being too complex.Same evaluation
2 extra hours per day on additional administrative tasks to address gaps in data recorded.An office manager, same evaluation
Three things, in order.
Care staff record at the point of care
Three taps on a phone in the corridor, with 48-pixel targets because staff wear gloves, and an offline queue for the dead spots. Nothing is written from memory at 14:00.
Verity reads the whole record back
It searches every note, observation, chart, incident, body map and MAR entry — not the last week, and not one record type at a time.
Every answer arrives with its evidence
Each claim carries a numbered citation that opens the exact record it came from. An inspector asking "how do you know?" is answered with a click.
Ask the question a manager actually asks.
"Summarise the last 30 days." "What has changed?" "Are there patterns management should review?" Verity answers from what is written down, and shows you where every word came from.
Three behaviours no other UK care platform has.
It abstains
"The record does not contain enough information to answer this" is a first-class output, rendered as calmly as an answer — and counted. A confidently wrong summary in a care record is a patient safety incident.
It shows contradictions
When two records disagree, both are shown, side by side. Averaging them or quietly picking one destroys exactly the evidence a safeguarding enquiry needs.
It refuses to diagnose
It will describe a pattern and stop. That boundary is what keeps the product outside medical device regulation, and it is enforced in the product rather than promised in a document.
It finds what no single person was looking at.
Overnight, Verity reads every record in the home and raises only what several independent sources point at — or where a record a protocol requires is simply missing.
Each finding is dismissible with a recorded reason. A manager showing they reviewed a finding and decided what to do about it is itself evidence of being well-led.
A different product for the person with gloves on.
Not the desktop screen squeezed into a phone. Three destinations, 48-pixel targets, one task per screen, the button in the thumb zone, and the offline queue permanently in the header rather than in a toast nobody sees.
The escalation rules live in the product, not in the carer's head. Answer "could she have hit her head?" on a resident who is anticoagulated, and it stops you before you submit anything.
It already speaks the language of an inspection.
Every instrument below is implemented with its real thresholds and its real escalation, held as reference data so a guideline change is a configuration edit rather than a release.
A twelve-minute walkthrough.
Every screen is live and clickable — nothing is behind a login. The three highlighted rows are the moments worth slowing down for.
See Verity answer a question no one in the home had put together.
Open the record first so the scale of it lands, then ask the question. It takes about four minutes and it is the whole argument.
What this is. A working, clickable prototype. Static HTML with no build step, no server and no external requests — it opens from a file and cannot break in a meeting. The retrieval behind Ask Verity is real: it searches generated records, the citations resolve to actual entries, and every figure on this page is computed from that corpus rather than typed in. The language layer is scripted so the demo is deterministic.
What it is not. There is no backend, no authentication and no live model. Every resident, staff member and home is fictional — nothing here is a real person or a real clinical record. Figures quoted from the NIHR evaluation and from sector research are real and are cited in README.md.