The first version of WellDash is two things: documents in and evidence out, and a record you can talk to. Send the care notes, incident records and letters your service already produces — from any system of record — and get typed evidence back. Then ask the record questions, in writing or out loud, and get answers from what is actually in it.
You can start on any rung and the evidence is real. What changes as you climb is how much of what actually happened survives into the record — and therefore what the record can later prove.
Send what you already produce — care notes, incident records, letters, existing plans, exports from any system of record. They are read into typed evidence against the right person, with the source document kept attached. You get retrospective evidence out of records that were previously only searchable by eye.
A carer says what happened and the words become a typed structured row rather than a note. And the record answers back: ask a question, in writing or out loud, and get an answer assembled from this person’s own recorded evidence and their authored plan.
The response dials — strategy, cue, assistance, manner — plus who performed the task and what the hands were doing, recorded at the moment of care. This is the rung where evidence stops being a description of what happened and becomes something that can be compared, ordered and argued from.
Most care software asks a service to switch its whole way of recording before it returns anything at all. That is the wrong order, and it is why so many good systems die in month three.
So the first version is deliberately narrow. Send in the records you already produce. Get typed, person-specific evidence back. Ask the record questions and have it answer from what it holds. Everything else in this section — the fade, the easier ask, the typed context, the body-first override — describes what becomes possible as the record gets richer, but none of it is a precondition for the first useful day.
The practical effect for a provider is immediate: the last two years of notes, letters and incident forms stop being a filing system and start being evidence about a person.
The second half of version one is conversation. It runs both ways, and the two directions do different jobs.
Each rung is genuinely useful, and each one is bounded by what the rung below could not hold. Saying so plainly is more useful than pretending the first rung is the whole product.
So the ladder is not an upgrade path invented for pricing. It is the honest statement that the questions a funding panel asks — prove it works, prove it is least restrictive, prove change is managed — can only be answered from the rung that recorded the answer.
And the honest comparator for all of it is not a rival product; it is the commissioned assessment and the hand-built panel submission that this evidence stands beside. That comparison has its own note: storage or evidence.
No. The first step is documents in and evidence out: you send the records you already produce — care notes, incident records, GP and clinic letters, existing plans, exports from whatever system you run — and they are read into the typed record. Nothing has to be switched off for that to be useful.
It is held as a source with its own identity, a targeted identity scrub runs over the text, and the content is read into typed evidence against the right person. The original stays attached, so every piece of typed evidence can be traced back to the document it came from.
Yes — that is the second half of the first version. You can ask in writing or out loud and get an answer assembled from this person’s own recorded evidence and their authored plan. Where the record is silent, the answer says so rather than filling the gap.
Fidelity. A document contains what somebody chose to write down after the event, in whatever detail the shift allowed. It cannot contain the cue rung that was used, who actually performed the task, or what the hands were doing — because nobody was asked. Retrospective text gives real evidence with real limits, and the limits are the reason the capture layer exists.
No. An ambient scribe turns speech into a note. Here the speech becomes a typed row in a behaviour grammar — form, parallel function hypotheses, the response configuration, who did it, before and after — with the carer’s own words preserved beside the tokens. The modality is the same; what the modality produces is not.
In the documents somebody has to produce anyway: the care plan, the behaviour-support plan, the emergency plans, the log book over any date range. They assemble from the same record the shift is running, and a field nobody recorded prints as a gap rather than a guess.
The demonstration shows capture, the typed grammar and the ranked responses on synthetic records.