Storage, or evidence.

Capture fidelity is the floor of everything

Everything downstream — every count, every comparison, every review — is bounded by what the capture step preserved. A dropdown loses the sentence. A form captures what the form designer imagined, which is a fixed set of things thought of in advance by someone who has never met this person.

Voice-first capture that types itself changes the economics of richness. The carer says what happened, in their own words, and the record ends up holding both the sentence and the typed pieces it becomes. Nobody pays for the richness with extra minutes, because the evidence was going to be spoken anyway — to a colleague at handover, if not to the record.

The test to apply to any product: ask what happens to a sentence that does not fit the form. In a storage-grade system it is discarded or squeezed. A fidelity-grade system keeps the sentence, types what it can, and flags what it cannot.

Teaching already happens — the record just loses it

Support staff teach behaviour swaps constantly, informally and often brilliantly. Someone works out that offering the headphones before the noise starts prevents the whole episode, and it works, and it stays in that person’s head until they move teams.

So the software’s job is not to make staff teach. It is to capture where the teaching already happened, transfer it to the team as a consistent palette of functional equivalents, and let the same swap be re-taught next shift, next building, next carer. Consistency is a palette problem, not a person problem — and blaming inconsistency on individual staff is the standard way to spend a training budget on a record-keeping failure.

Evidence for a panel is computed, or it is assembled by hand

For the highest-need funded placements, a panel asks three things: prove the intervention works, prove the least-restrictive case, prove change is being managed. There are exactly two ways to answer.

• Computed. The dated dechallenge, the outcomes filed by function, the recorded fade, and any proposed plan change carrying the sources that prompted it — accumulating as the work happens, because the shift recorded the work.
• Hand-assembled. A manager, the night before, from notes, at overtime rates, reconstructing six months of care into a narrative that must then be believed.

The regulatory frame is part of the same substrate rather than a separate exercise: a plan step can carry the CQC quality references it satisfies and the person’s own I-statement outcome it serves. Those references are stamped onto the record when the step is completed, and the I-statement prints on the plan document as the outcome the step exists for — so the quality framing is a property of the care rather than a mapping exercise performed once a year.

Change management is structural, not exhortation

Every provider has watched a good initiative die of adoption. The usual diagnosis is culture, and the usual prescription is training. Both are usually wrong: people stop using a system when using it correctly produces nothing they can see.

When change lands as data — when the plan visibly re-orders, with the reason attached, on the screen the carer is already looking at — adoption stops being a training problem and becomes a by-product of use. The person recording watches their own recording change what the next shift is asked to do. That is a structural property of the system, not a cultural achievement of the manager.

The price difference is the evidence difference

There are two price classes in this market and they buy two different things. At the bottom, around £7 per person per month, you are buying form-filling: storage with templates, competently done. At the top, around £70, you are buying a record that argues — capture fidelity, computed evidence, engineered change.

Stated as a property rather than a boast: the step from one to the other is the step from storing what happened to being able to prove what worked. For a provider with one highly funded person going to panel, the difference between the two classes is one successful funding outcome — which is the arithmetic that actually decides these purchases, whatever the feature comparison says.

What this evidence is actually standing beside

The honest comparator for computed care evidence is not a rival piece of software. It is the commissioned independent assessment — a functional or PBS assessment bought in when a placement is under pressure — and the panel submission built by hand to go with it.

That is a four-figure exercise. As a typical UK market range rather than a published rate, a commissioned functional or PBS assessment sits somewhere around £1,500 to £3,000, and the fee is the smaller half: briefing it, hosting the observation visits, assembling the historical evidence it asks for and acting on the report costs days of senior practitioner and manager time. Treat the range as an estimate — no UK body publishes a standard fee for this work, and we would rather say so than dress a market impression as a citation.

The structural difference is not price, though. It is sampling. A commissioned assessment observes the person for hours, interviews the people who were there, and writes up a moment. Computed evidence is derived from the person’s whole recorded history and is current on the day it is read. One is a photograph taken by an expert; the other is the accumulated record of the life, read with a method.

This is where the four pillars stop being features and become the argument. Each one is a reason computed evidence can answer something a sampled assessment structurally cannot:

• The fade — an assessment describes the support someone needs today. The record holds the rung used on every occasion, dated, so the direction of travel is evidence rather than recollection. Read the note.
• Functional equivalence — an assessment recommends a replacement behaviour. The record shows whether the replacement actually earned the same wage afterwards, on the person’s own occasions. Read the note.
• The typed context — an assessment collects antecedents from interviews and a few visits. The record accumulates the background states as they occurred, including the body clocks nobody would think to mention in an interview. Read the note.
• Body first — a behavioural assessment is a behavioural instrument. The record enforces the physical-cause order continuously and holds the dated dechallenge, which is the single strongest piece of evidence in the whole file and is produced by care rather than by assessment. Read the note.

What we looked for, and what we did not find

In an August 2026 timestamped survey of UK care-management platforms, UK PBS-specific tools and US applied-behaviour platforms, we found no deployed system that indexes behaviour evidence by function, computes functional equivalence from outcomes, or dates the fade of assistance on real-life response configurations.

Two honest qualifications, because a claim without them is not worth making. First, voice capture itself is commoditising quickly: ambient voice tools arrived across UK care software through 2025 and 2026 and will keep arriving, so the modality is not the differentiator — the typed grammar the speech becomes is. Second, the nearest computational work we found sits in applied-behaviour tooling that produces summary statistics over pre-categorised entries; it does not touch narrative, response configuration, or per-person outcome ordering. We name the scope of the search rather than claiming the world.

What this puts in front of a commissioner

The whole point of the five premises is what happens at the end: somebody has to write the report, and somebody has to decide on it.

• The care plan and the behaviour-support plan print from the same substrate the shift is running — the document and the practice are the same object rather than two versions of it.
• The evidence a reviewer asks for was recorded as the care happened, with dates, rather than reconstructed under time pressure by someone who was not there.
• Regulatory references and the person’s own stated outcomes travel with the record, so quality framing is a property of the care rather than an annual paperwork exercise.
• Every gap is visible as a gap. A field nobody recorded prints as a gap, never as a guess — which is the property that makes the rest of it trustworthy.

Questions, answered the way we answer them

The mechanisms behind the premises

• Documents in, evidence out — where a service starts, and the fidelity ladder the premises are measured on.
• The four pillars — the grammar the premises describe.
• Fifty years to find the function — why function-indexed evidence is the unit.
• The fade — what a dated fade looks like in a record.