Industries

AI systems by industry

The plumbing is often similar from one sector to the next. The constraints almost never are. What changes between a hotel group and an NHS trust isn't the technology — it's the regulation, the data sensitivity, and what happens when the system gets something wrong.

Does the industry actually change how an AI system is built?

Partly. The underlying components — document extraction, retrieval, drafting, human review — recur across almost every sector. What changes is the constraint set: which regulator is watching, how sensitive the data is, where the system has to run, and which decisions a human must keep.

That distinction matters when you're buying. An agency that only talks about the technology will build you something that works in a demo and fails an audit. An agency that only talks about compliance will sell you a policy document and no working software.

In practice, the sector determines four things: the deployment model (cloud, private tenancy, on-premise, or fully air-gapped), the evidence you need to produce for a regulator, auditor or board, the escalation design — what the system does when it isn't confident — and the hard limits on what it must never decide by itself.

How the nine sectors differ

Typical first builds, the constraint that shapes them, and how long they take. Timelines are from signed scope to live, and assume one workflow rather than a whole-business rollout.

Sector Most common first build Hardest constraint Typical deployment Timeline
Legal services Matter intake, conflict checking and disclosure Privilege and client confidentiality Private tenancy, matter-level separation 6–8 weeks
Healthcare & care Administrative and coordination workflows The line between admin and clinical judgement On-premise or air-gapped 8–12 weeks
Logistics & operations Document processing and exception handling 24/7 running; failures cost money fast Cloud, integrated with TMS/WMS 4–8 weeks
Financial services Client onboarding and document handling Explainability and an evidenced audit trail Private tenancy or on-premise 8–12 weeks
Hotels & hospitality Guest enquiry handling and booking admin Peaks, seasonality and high staff turnover Cloud, integrated with the PMS 4–6 weeks
Public sector & charity Correspondence triage and reporting Transparency, auditability, procurement On-premise or air-gapped 8–12 weeks
Property & estates Enquiry triage and compliance records Fair treatment of applicants and tenants Cloud, integrated with the CRM 4–6 weeks
Education & training Administrative load and content preparation Children's data and safeguarding Private tenancy, UK-hosted 6–10 weeks
Retail & e-commerce Customer messaging and product data Peak-season volume; brand voice at scale Cloud, integrated with the storefront 4–6 weeks

Which constraint drives your build — data sensitivity, regulation, or volume?

Most projects are shaped by one dominant constraint. Identifying which one you have is the fastest way to know what your build will cost, where it will run, and how long it will take.

If the constraint is data sensitivity

Personal, clinical, financial or otherwise confidential data that cannot leave your control. Expect on-premise or air-gapped deployment, a Data Protection Impact Assessment before go-live, and a longer timeline.

If the constraint is regulation

A regulator who will ask how a decision was reached. Expect explainability requirements, immutable logging, documented human oversight, and EU AI Act risk classification as a first step.

If the constraint is volume

High throughput, seasonal peaks, or work running around the clock. Expect the design effort to go into exception handling and escalation rather than the happy path.

Sector pages, in detail

Each one covers what we'd build, what we won't, and the regulation that shapes it. We're publishing the remaining sectors from the table above over the coming weeks — if yours isn't up yet, just ask us.

What if your industry isn't listed?

It's usually not a problem. The nine sectors above are the ones we're asked about most often, not the limit of what we work on. The questions we'd ask you are the same regardless of sector: what the work is, how sensitive the data is, who is accountable for the output, and what happens when the system is wrong.

If you can answer those four, we can scope a build — whether or not your industry has a page here. Tell us what the work is and we'll tell you honestly whether AI is the right tool for it.

Start with the problem, not the technology.

Tell us what's slow, repetitive, and costing more than it should. We'll tell you whether AI can fix it — and if it can't, we'll say so.

Let's talk

Last reviewed: February 2026