AI readiness for healthcare companiesEvery clinic keeps good records. Just not the same ones.
Many healthcare groups are built by acquisition, and every site arrives with its own practice management system, its own codes and its own spreadsheets. AI for scheduling, billing and patient communications needs one patient, one appointment and one fee to mean the same thing across the group.
Starts with a 30-minute call. Commits you to nothing.
The AI on the table
The AI plays proposed in healthcare.
What vendors and plans put forward in healthcare, what each play needs from the data, and the gap that holds it back. In a real Read, each call rests on your Truth Score™ (a score out of 100 for how far your data can be trusted) and the records behind it. The calls here are an illustrative read, not a benchmark.
-
Appointment and capacity forecasting
“Predict no-shows and fill the diary”
What demand forecasting needsNeeds work- Needs
- Appointment history coded the same way at every site
- Held back by
- Appointment types named differently per site
- Blocked
- Needs
- Treatments that match fees and invoices
- Held back by
- Treatments recorded with no matching invoice or claim
-
Referral letter extraction
“Read referral letters into the patient record”
What document intelligence needsNeeds work- Needs
- Extracted details that match one patient record
- Held back by
- The same patient registered at two sites
-
Patient communications agent
“Answer booking and balance queries automatically”
What customer service automation needsNeeds work- Needs
- Patients tied to bookings and balances
- Held back by
- Balances that differ between site and finance
-
Group cash forecasting
“A 13-week cash view across every site”
What 13-week cash flow forecasting needsNeeds work- Needs
- Site takings that reconcile to the bank and ledger
- Held back by
- Insurer and self-pay receipts not allocated to invoices
Illustrative example, not a client. Your calls depend on your records.
The systems estate
The systems behind the plays.
Written for clinic and practice groups (dental, veterinary, physiotherapy, diagnostics), care providers and healthcare services companies, many built by acquisition, with a practice management system at each site. Each system is built to be right about one thing. None was built to agree with the others, and the plays above all run across more than one.
-
Practice management
Patients, appointments, treatments
A system per site, sometimes several products
-
Finance
Ledger, fees, site takings
-
HR, rostering and payroll
Clinicians, rotas, pay
-
Referrer CRM
Referrers, campaigns, enquiries
-
Payments and claims
Card takings, insurer remittances
Terminals and insurer portals
-
Spreadsheets
Site KPI packs, group reporting
The monthly pack
Typical, not exhaustive. Anything that exports to CSV or Excel can be included.
The gaps
Every system looks clean on its own. The truth is in the gaps between them.
One record traced across each site’s practice system, finance and the rota: each system right about its own part, and the three still disagree. Then the gaps that hold the plays back. Which of them exist in your data is what a Truth Score Read finds out.
Where the plays sink
-
Site ↔ Site
The same patient registered at two sites, with different details
SinksPatient communications and capacity planning
-
Practice ↔ Finance
Treatments completed with no matching invoice or claim
SinksBilling automation and cash forecasting
-
Rostering ↔ Practice
Clinicians in the diary who aren’t on payroll, or the reverse
SinksCapacity forecasting
-
Spreadsheets ↔ Finance
Site KPI packs the ledger can’t reproduce
SinksGroup reporting and the investor plan
Gaps to test in the sector, not client findings.
Regulation
The rules that shape which plays you can run.
Kept short and honest. This is not legal advice: we evidence what your data can support; your counsel decides what the rules require.
GDPR: health is special category data
We agree exports in advance, under NDA and a data processing agreement, and can work from pseudonymised extracts where the use case allows. We keep quality metrics, not patient records.
The EU AI Act and medical devices
AI that informs diagnosis or treatment can be a medical device and high-risk. The plays on this page are operational; clinical AI needs its own regulatory path, which we don’t advise on.
National health records rules
Retention, patient access and professional record-keeping duties vary by country and regulator. Your data protection officer sets the limits; we design the exports within them.
Your plan, tested
Got a plan that names these plays?
Send it with your exports. A Truth Score Read tests the AI in it against your own records, within 5 working days of complete exports. Go further with an AI Investment Review: every bet ranked, your plan tested play by play, and a roadmap to hand to your own team.
- 01
Send the plan, if there is one
The business plan, growth strategy, value creation plan or the vendor proposal. We pull out the AI plays in it and the targets that lean on them.
- 02
Send exports from two or more systems
Usually practice management and finance to start. Spreadsheets count. Under NDA, using the field list we send.
- 03
Test each play against the records
Each play mapped to the data it needs. Our engine matches the customers, people, deals and invoices across your exports to see whether that data is there and agrees, and every red flag is checked by hand against sample records.
- 04
Get the verdict, play by play
A signed memo: your Truth Score, a Leverability read marking each play Deploy now, Needs work or Blocked with the reason in your own records, and a verdict: sign, fix first or walk away.
What you get back
See what comes back: a real sample memo.
One fictional B2B software company, Example Co: 72,837 records in 7 exports from its CRM, finance and HR systems, run through the same engine and the same checks as a paid Truth Score Read. Each system passes its own checks. Matched across systems, only 13% of deals tie to an invoice.
Not a software company? The sample is one, but the method doesn’t change; only the systems do. For a healthcare business the same match runs across each site’s practice system, finance and the rota, and every red flag is checked by hand the same way.
Questions
AI in healthcare, answered plainly.
Anything else, ask on the 30-minute call. It commits you to nothing.
Book the callDo you need patient records?
Not in identifiable form, as a rule. For operational plays we can work from pseudonymised extracts: appointment, treatment and invoice records with patient identifiers replaced by a consistent code, so records still match across systems.
We have several practice management systems after acquisitions. Is that a problem?
It is a common reason group-level AI stalls, and exactly what cross-system matching is for. We compare the sites’ exports with each other and with finance, so you can see where codes, patients and fees diverge.
Do you assess clinical AI?
No. We test whether operational data (scheduling, billing, finance, workforce) can support the AI being proposed. Clinical decision support needs clinical and regulatory review that sits outside our scope.
Our investor wants a group-wide AI plan. Where do we start?
With the plan itself. Send it with exports from two or more systems, and we test the plays in it against the data. If the sites run different codes, the memo says so plainly: one set of codes comes before any group-wide AI is credible.
Before healthcare AI gets funded, find out what your data can carry.
Exports from two or more systems, plus the plan or proposal if there is one. A signed verdict, play by play, within 5 working days of complete exports. Starts with a 30-minute call. Commits you to nothing.
Starts with a 30-minute call. Commits you to nothing.