The Hardest Problem in Care Tech Is Not the AI. It Is the Buyer.

A colleague asked me a question recently that cut through everything I thought I knew about this product.

I was talking through a care-tech idea I have been shaping around children’s residential care.

The idea is simple enough on the surface. A digital platform for children’s homes that helps track young people’s daily emotional wellbeing, gather feedback, structure key-working sessions, and flag safeguarding concerns early using pattern detection.

Not AI replacing staff.

Not a shiny dashboard for the sake of it.

A practical system that helps staff, managers and responsible individuals see what is happening across a home before things are missed, buried in logs, or remembered only by the person who happened to be on shift.

My instinct was that the problem was obvious. Children’s homes are dealing with complex young people, high emotional risk, fragmented records, regulatory pressure, and very little time. If technology can help surface patterns, improve recording, and make feedback easier to capture, surely there is a strong case for it.

Then my colleague asked the question that cut through the excitement:

Who is the buyer?

Not who benefits. Not who needs it. Who actually pays?

That question is uncomfortable because in children’s social care, the person who benefits from better systems is often not the person who controls the budget.

The young person benefits from being heard earlier. The staff member benefits from clearer structure. The manager benefits from better oversight. The regulator benefits from better evidence. The council benefits if placement quality improves. The provider benefits if the home is safer and better run.

But who signs the contract?

That is where care tech gets complicated.

Councils have the problem, but not always the appetite

Local authorities are under massive pressure. They are responsible for children in care, but many do not directly provide enough residential placements themselves. They commission private providers, negotiate placements, manage budgets, deal with safeguarding concerns, and try to find suitable homes in a market that is already stretched.

From a pure problem-solving point of view, councils should care deeply about better visibility. They should want to know:

Which placements are becoming unstable? Which young people are repeatedly reporting anxiety, fear, isolation, or conflict? Which providers are recording meaningful key work — and which ones are just producing paperwork? Where are safeguarding concerns emerging across multiple homes? Which homes are reactive, and which ones are genuinely learning?

But councils also operate under brutal financial pressure. If a council can solve a problem by asking for more staff time, another review meeting, another spreadsheet, or another reporting template, it may choose that route over paying for a new technology solution.

That does not mean the technology has no value. It means the buying logic is different.

Councils may not buy “better AI.” They may buy better commissioning oversight, reduced placement breakdowns, clearer provider accountability, early safeguarding visibility, evidence that helps them challenge poor-quality provision, and a way to understand what is happening across placements they do not directly control.

That is a very different product story.

Private providers may be the first buyer, but not always for the reason you hope

At first, I thought the natural buyer would be private children’s home providers. They run the homes. They employ the staff. They need to evidence care quality. They face Ofsted inspections, Reg 44 visits, local authority scrutiny, complaints, incidents and safeguarding escalations.

Many already use digital systems, but in my experience some of those systems still feel behind the reality of the work. They record information, but they do not always help people understand what is changing for the young person.

That gap matters. A good care-tech product should not just store daily logs. It should help answer questions like:

Has this young person’s emotional presentation changed over the last two weeks? Are we following up on key-working actions? Are the same themes appearing across incidents, feedback and staff observations? Is a young person repeatedly saying something in different ways that staff have not yet connected? Are managers seeing patterns early enough?

This is where AI can help — but only if it is used carefully. The value is not “let the machine decide.” The value is “help humans see the pattern sooner.”

But the private provider market has its own tension. Some providers genuinely want better systems because they care about quality, evidence and outcomes. Others may only care if the system protects their rating, reduces admin, helps them win placements, or reduces commercial risk.

That sounds cynical. But it matters.

If the sector contains operators who see children’s homes mainly as an investment opportunity, then selling care tech as “better outcomes for young people” may not be enough. The product also has to speak the language of compliance, inspection readiness, local authority confidence, incident reduction, placement stability, reputation protection, and operational efficiency.

That is not selling out. That is understanding the market without lying to yourself.

The market problem is harder to ignore than it used to be

The Financial Times published a piece titled “How England’s vulnerable children became a gold mine for investors” that was difficult to read. It is behind a paywall — but the Ofsted Chief Inspector said something similar in public.

In his 2024/25 annual report, Sir Martyn Oliver stated:

“Profit motive is increasingly dictating the location and ownership of children’s homes. As a society, we are failing these children.”

That is not a journalist’s opinion. That is the regulator.

The picture it describes is uncomfortable. Government spending on children’s homes has risen sharply. Private providers dominate the market. Children’s residential care has become attractive to investors and new entrants with limited care experience. Some areas have far more children’s home places than local demand. Others have very few. There are concerns about unregistered placements, poor quality, and children being moved far from family, school and community.

That kind of evidence matters because it shows the problem is not just “care homes need better software.” The market itself is distorted. There is money in the system, but not always in the right places. There is demand, but not always proper quality. There is regulation, but not always enough visibility.

And there are children whose lived experience can disappear behind placement fees, provider language, inspection outcomes and commissioning pressure.

So when I think about my care-tech product now, I do not see it as just a product question. I see it as a system question.

Where does technology sit in a market where vulnerable children can become revenue lines? How do you build tools that support care without becoming part of the same machinery that turns care into margin? How do you sell to providers without designing only for provider convenience? How do you support councils without adding another procurement burden? How do you use AI without creating surveillance, risk scoring, or lazy decision-making?

These are not small questions.

The product cannot just be “nice to have”

A care-tech product in this space cannot be positioned as a nice dashboard. Nice dashboards die. Staff do not need more screens. Managers do not need another system that creates more work. Councils do not need another supplier promising transformation while adding procurement friction.

The product has to sit directly inside the pain.

For providers, that pain might be Ofsted evidence, Reg 44 preparation, key-work tracking, safeguarding audit trails, staff recording quality, reducing missed follow-ups, and proving the home is responsive to young people’s views.

For councils, the pain might be poor visibility across commissioned placements, difficulty comparing provider quality, high-cost placement breakdowns, children being moved repeatedly, and weak evidence when challenging providers.

For staff, the pain is simpler. Too much recording, not enough time, information scattered everywhere, important patterns living in people’s heads, and key work happening but not always being evidenced properly.

For young people, the pain is the most important. Not feeling heard. Saying something once and nothing changing. Repeating the same concern to different adults. Feedback being treated as paperwork. Emotional changes being noticed too late.

The product has to prove it can reduce those pains. Not theoretically. Practically.

My current view: start with providers, design for accountability

If I had to choose a realistic entry point, I would still start with private providers — but not with a broad “AI care platform” pitch. I would start narrower.

A lightweight system for children’s homes that helps them capture young people’s feedback through QR forms, structure key-working sessions, track follow-up actions, flag safeguarding language or repeated concerns, and produce clear oversight reports for managers, Reg 44 visitors, Ofsted and local authorities.

That is a much clearer wedge. It does not try to replace existing care management systems immediately. It sits on top of poor or fragmented recording and adds value where the current process is weakest — feedback, themes, follow-up, emotional patterns and evidence.

Over time, if the product proves itself, it can become more powerful. But the first version should answer a basic question:

Can this help a home see, evidence and respond to young people’s needs better than it does today?

If the answer is yes, the commercial story becomes stronger. Not because the AI is clever. Because the operational value is clear.

AI must remain the assistant, not the authority

There is a dangerous version of this product. The dangerous version turns children into risk scores. It scans their records, assigns a number, and gives managers the illusion of control.

I do not want to build that.

The safer version is more modest.

AI can help identify language that may need review. AI can group repeated themes. AI can summarize long records for staff who are drowning in information. AI can highlight that a young person has mentioned feeling unsafe several times in different contexts. AI can help managers spot what they may have missed.

But AI should not decide whether a child is safe. It should not replace professional judgement. It should not generate fake therapeutic insight. It should not become a shortcut for poor staffing, poor supervision or poor care.

In this space, AI should behave less like a judge and more like a smoke alarm.

It does not tell you the whole story. It tells you to look properly.

The real challenge is trust

The more I think about this product, the more I realize the technical build may not be the hardest part. Yes, the architecture matters. The platform needs secure data handling, proper access control, audit logs, backups, monitoring, and careful AI boundaries.

But the deeper challenge is trust.

Can staff trust that the system helps rather than polices them? Can managers trust that the data is accurate enough to act on? Can councils trust that the provider is not just generating polished reports? Can young people trust that their feedback will not disappear into another adult system? Can regulators trust that AI is not being used to hide weak practice behind clever language?

That is the hill this product has to climb.

Where this leaves me

The conversation with my colleague made me less naive. That is a good thing.

It forced me to separate three questions:

Is there a real problem? Yes. Children’s residential care has serious issues around feedback, evidence, visibility, placement quality and safeguarding oversight.

Can technology help? Yes — but only if it is grounded in real care practice, not fantasy product thinking.

Is there a clear buyer? That is the hard part. The buyer is not automatically the person with the biggest need. The buyer is the person who can connect the product to money, risk, regulation, evidence, reputation or operational pressure.

That means the product has to be designed with both ethics and commercial reality in mind. If it only speaks the language of care, it may not sell. If it only speaks the language of efficiency, it becomes part of the problem.

The balance is the work.

For now, I still believe there is a place for carefully designed care tech in children’s homes. But I am clearer now that the opportunity is not simply “AI for social care.”

The opportunity is building reliable, ethical, evidence-focused systems that help the right people see the right risks earlier — without turning vulnerable children into data products.

That is a harder pitch. But it is a better one.

And it is the only one I can make honestly.

Victory Anusie is a Site Reliability Engineer, DevOps practitioner and frontline residential childcare worker building myCareSignals — a child-pathway intelligence platform for children’s residential care homes in England.