Domiciliary Care Software UK: What the AI Features Actually Do

Of the three best-known options, Birdie has the only live AI doing real work: SmartPlans turns a recorded assessment into suggested care-plan answers linked to the transcript, never drafts high-risk clinical questions, and is billed per recording hour (£3.50 after three free hours). CareLineLive names no AI, and Nourish's AI product, Confidence, is announced as upcoming.

That matters because most of what a registered manager needs for inspection comes from ordinary digital records, not AI: medication records, visit times, audits and actions. Choose the platform on rostering, eMAR and price per hour or per carer first, then treat AI assessment drafting as a supervisor-time saving you can measure.

Follow me on Instagram@sagnikteaches

Where the office hours go in a home care agency

A home care agency coordinates carers travelling between people's homes, often several hundred care hours a week. Field supervisors carry out initial assessments and reviews covering mobility and transfers, skin integrity, eating and drinking, medication, continence, memory and the home environment. Coordinators rebuild the rota every time someone is ill or a visit runs over. Office staff chase medication records and audit them.

Connect on LinkedInSagnik Bhattacharya

The assessment is where AI can help most. A supervisor may spend over an hour in the person's home taking notes and then two or more hours typing them into templates. Across a month of new clients and six-monthly reviews, that typing is a large share of a skilled supervisor's week, time that could go into spot checks, carer support and safeguarding.

Subscribe on YouTube@sagnikteaches

Inspection adds a second pressure. CQC looks for evidence that care is safe, effective, caring, responsive and well-led. Paper files scattered across homes make that evidence slow to assemble. Digital records do not create good care, but they make it far easier to show.

Separating working tools from sales claims

The rule that separates useful AI from risky AI in social care is that every suggestion must be visible and traceable. If software drafts an answer about someone's morning routine, a supervisor should be able to click it and see what the person or their family actually said. Birdie's SmartPlans page describes exactly this: each answer is linked back to the part of the conversation it came from. A vendor that cannot show you the source of a suggestion is asking you to sign something you cannot check.

Rotas are a separate problem; see building a domiciliary care rota and finding its gaps. For the quality checks that follow visits, domiciliary care spot check templates and action lists covers the paperwork.

Birdie, CareLineLive and Nourish compared

From each vendor's own pages, read on 4 November 2026. Prices are shown excluding tax and are reviewed by the vendors; confirm them before you sign.

Birdie's SmartPlans works like this, according to its page: you press record in the Birdie app before the conversation starts, the recording is transcribed and matched to the assessment questions, and the audio and transcript are stored in the person's record for replay. Questions are handled in three ways. Standard, lower-risk questions get a suggested answer that the supervisor reviews and confirms. Complex judgements get the relevant quotes highlighted, and the supervisor writes the answer. High-risk clinical questions get no suggestion at all. Birdie also says the data stays within its platform and is not used to train AI models. The Birdie pricing page shows that SmartTools sit outside the plan price, so budget for them separately.

CareLineLive's case is simple pricing. The CareLineLive pricing page charges per carer per month, with unlimited office licences included free, and a £120 monthly minimum that the page says covers 15 carers on eRoster, 7 on bring-your-own-device or 3 on managed handsets. Its home page names real-time call monitoring with alerts, electronic visit records with eMAR, and a Care Circle Portal for families. It makes no AI claim, which is honest, and which means any AI you want would come from elsewhere.

Nourish brings residential-care experience to home care through Better Care at Home, which its page says includes the rostering, invoicing and time-off functions of Nourish's separate rostering product, an offline mode for remote teams and a Family Portal. Its AI page describes Nourish Confidence as "our next AI Product", with AI-assisted care plan reviews; treat it as upcoming until you have seen it working in a demo. For a wider plan of where to start with AI in a care service, see a registered manager's 90-day AI implementation plan.

Demo questions that show what is live and what is roadmap

Sales demos run on prepared data. These questions move the demo onto your ground, and the answers tell you whether a feature exists today:

• For any AI assessment tool: "Record a five-minute mock review with me now and show me the suggested answers." Then click two suggestions and check that each opens the quote it came from. Ask which question types never get a suggestion, and whether you can change that list.
• For pricing: "What would I pay at 1,200 hours a week, including AI recording charges, devices and support?" Ask for it in writing, with the annual review terms.
• For offline use: "Put the carer app in airplane mode, log a visit with a medication, then reconnect." Watch what happens to the eMAR entry and its time stamp.
• For anything described as coming soon: "Is there a release date, and will it cost extra?" Price it at nothing until it is in your contract.
• For data: "Where are recordings stored, how long are they kept, and are they used to train models?" Birdie's SmartPlans page covers where data is held and training; get all three answers in writing from every vendor.

Take a supervisor and a coordinator to every demo. They will spot within minutes whether the assessment screens and the rota board match how your service actually runs.

Pricing a 1,200-hour agency on both models

An illustration with assumed figures. Picture an independent agency delivering about 1,200 care hours a week with 40 carers and 85 clients. One field supervisor does around 18 assessments and reviews a month, each with roughly 90 minutes of conversation and 2.5 hours of typing afterwards. Two coordinators run the rota.

CareLineLive cost. 40 carers on bring-your-own-device at £19.70 is £788 a month. On eRoster only it would be £324. Office staff are free either way.

Birdie cost. The plan price depends on scheduled hours and is quoted, so the agency needs a quote for 1,200 hours a week. The AI part can be estimated now: 18 assessments at 1.5 recording hours is 27 hours a month. On pay-for-use, the page offers three free hours and then £3.50 an hour, so about 24 x £3.50 = £84. On a subscription from £2.75 an hour, about 27 x £2.75 = £74.

The comparison that decides it is not CareLineLive against Birdie's AI. It is total platform cost (Birdie's quote against £324 to £788 a month for CareLineLive) plus the value of the supervisor hours. If Birdie's quote for 1,200 hours is close to CareLineLive's per-carer total, the SmartPlans saving tips it. If it is far higher, the agency may do better on per-carer pricing and accept slower assessment write-ups. The 40-minute review time is the assumption to test in a pilot: if supervisors end up rewriting most suggestions, the saving shrinks.

Clear carer notes matter as much as assessments; the tutorial on reviewing daily care notes with AI shows how to check wording without changing facts.

Turning digital records into CQC evidence

CQC assesses providers against five key questions: safe, effective, caring, responsive and well-led. In 2026 it also consulted on new sector-specific assessment frameworks, including one for adult social care (its provider bulletin asked for feedback by 12 June), so check the current framework on the CQC website before you label evidence with its wording. The evidence itself does not change much:

The strongest evidence is a trail, not a dashboard: a concern noticed, recorded, escalated and closed. For practice runs before the real thing, see summarising mock inspection findings with AI.

Five situations that show where AI helps and where it must stop

1. A high-risk transfer question in an assessment

During a review for a client recovering from a stroke, the supervisor asks the daughter about standing. She says: "Mum usually leans on me to stand, but sometimes her knees go and she slides to the floor."

A general AI model asked to "write the care plan" might produce "Assist with standing transfers using a walking frame", which could be dangerous for someone whose legs give way without warning. Under Birdie's model, moving and handling sits with the questions the supervisor writes. An illustrative screen:

MOVING AND HANDLING (supervisor to complete; no AI suggestion)
Relevant quote from the recording:
  Daughter: "sometimes her knees go and she slides to the floor"

SUPERVISOR'S ENTRY
Unpredictable leg weakness when standing, following a stroke. Not to be stood by one carer. Two carers for all bed-to-chair transfers, using the equipment specified in the occupational therapy assessment. Moving and handling assessment and sling sizing booked for Thursday.

The software's job was to surface the quote. The decision stayed with the qualified person.

2. Checking travel time on an exported rota

Rotas fail when every gap between visits is a flat 15 minutes regardless of distance or the school run. Whatever system you use, a coordinator can export a week's rota with visit IDs and areas only (no names or addresses) and ask a general assistant to flag tight gaps:

Here is a rota extract: carer ID, visit ID, area, start, end.
For each carer, list consecutive visits where the gap is under 20 minutes and the areas differ.
Output: carer ID, the two visit IDs, the gap in minutes.

Illustrative output: "Carer 07: visits 112 to 118, areas differ, gap 15 minutes." The coordinator then checks the real journey time and either moves the second visit or reassigns it to a carer finishing nearby. The AI only finds candidates; the coordinator knows the roads.

3. Rewording a carer's note without changing the facts

Carers often type notes quickly in the car between calls, and families may read them through a portal. A before-and-after, illustrative:

CARER'S NOTE
"Client was uncooperative and refused to let me wash her legs. She shouted at me to get out."

AI-SUGGESTED REWORDING
"Offered personal care to [client's preferred name]. She was distressed and firmly declined having her lower legs washed today, saying she preferred to rest. Respected her choice, made sure she was warm and comfortable with her morning tea, and told the office so the lunchtime carer could offer again."

The rewording is kinder, but read it again. "Saying she preferred to rest", "warm and comfortable with her morning tea" and "told the office" were never in the carer's note. If they did not happen, the note is now false. The carer accepts a rewording only after deleting anything they did not see or do, and the office checks a sample of reworded notes each week against what carers say in supervision.

4. Low fluid intake across one day's visits

Where a system records fluid targets and intake at each visit, a simple rule can raise an alert. An illustrative day for a client living with dementia:

The 16:30 alert lets the supervisor call the evening carer, ask them to encourage drinks and decide whether to contact the GP or community nursing team. Ask each vendor whether its system supports fluid targets and alerts like this, and who receives them; it is a rule, not AI, and it is worth more than most AI features.

5. A visit check-in that looks wrong

Call monitoring records when carers check in and out. Rural homes with long drives and blocks of flats can produce check-ins that look off-site. An illustrative exception log:

VISIT EXCEPTION
Visit ID 2231 | Scheduled 07:30 | Carer checked in 07:29
Location recorded about 120 m from the property (no signal at the door)
Action: coordinator phones carer, confirms arrival, notes reason on the visit
Outcome: visit confirmed; property flagged for an alternative check-in method

Ask vendors how check-in works when GPS is poor (a tag in the home, a phone call, a manual override with a reason) and whether every override is recorded. Inspectors and commissioners will ask the same question.

Switching systems: a five-step plan for the registered manager

1. Do the pricing sum for your size. Total care hours a week and active carers. Get Birdie's hours-based quote, work out CareLineLive's per-carer total, and ask Nourish for a quote, then add any AI recording charges.
2. Tell staff and clients what will change. Explain to supervisors that recording replaces evening typing. Give clients and families a plain explanation of why assessments are recorded and how to say no; those who decline get a typed assessment.
3. Decide devices and test signal. Managed phones or carers' own devices with security settings. Ask each vendor to demonstrate a visit logged with no signal and then synchronised.
4. Pilot AI drafting on five reviews. Your lead supervisor records five routine reviews. Count suggestions accepted, edited and rejected, and confirm that high-risk questions such as transfers and medication were left for the supervisor.
5. Build the evidence trail. Set up a folder or report set that pulls eMAR exceptions, punctuality, audits and actions by month, labelled against CQC's current framework.

After three months, compare supervisor write-up time, late-visit figures and the number of open audit actions with the three months before. Those three numbers tell you whether the software, and its AI, is earning its fee.

Domiciliary care software and AI questions

Can AI write a domiciliary care plan on its own?

No, and the tools that work do not try. Birdie's SmartPlans suggests answers only to standard, lower-risk assessment questions, links each suggestion to the part of the recorded conversation it came from, and never suggests answers to high-risk clinical questions. A supervisor reviews and confirms every answer before anything is saved to the plan.

How is domiciliary care software priced?

Two models are common. Birdie prices plans by the hours of care you schedule, from £200 a month, with its AI assessments billed separately per recording hour. CareLineLive charges per carer per month (£8.10 to £45 depending on the device option) with a £120 monthly minimum and free office users. Nourish quotes on request.

Does AI help with CQC inspections?

Indirectly. Inspectors look for evidence that care is safe, effective, caring, responsive and well-led, and digital records make that evidence easier to produce: medication records, visit times, care plans that quote the person's own words, audits and actions. CQC consulted on new assessment frameworks in 2026, so check its current guidance before mapping evidence to its wording.

What happens when a carer has no mobile signal?

It depends on the app. Nourish says its Better Care at Home app has an offline mode so carers can log care without a connection. The CareLineLive and Birdie pages read for this tutorial do not say, so ask each vendor to demonstrate a visit logged in airplane mode and then synchronised.

Further reads

• Care Home Audit Checklist: A Monthly Round Logged and Actioned With AI — Monthly quality audits and tracking the actions they raise.
• Care Home Mandatory Training Tracker: Renewal Dates and AI Overdue Alerts — Training compliance tracking with renewal reminders.
• Care Home Medication Audit: A Monthly Template and AI Action Log — Medication audit logs that hold up in an inspection.
• CQC Statement of Purpose Template: Draft Yours With AI and Check It — Drafting and updating a statement of purpose with AI.
• Can AI Write a Care Home Incident Report? Template and Checks — Consistent incident and safeguarding reports across the service.
• How Can AI Help in Care Homes? Ten Back-Office Uses — Back-office jobs in social care where automation helps.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Birdie SmartPlans and pricing pages; CareLineLive home and pricing pages; Nourish home, AI and Better Care at Home pages; CQC provider bulletin on draft assessment frameworks. All read 4 November 2026.