AI by Industry

Most AI advice is written as though every business were the same business. It is not. A marketing team, a dental practice, and an estate agency can all use the same chat assistant and still get completely different value out of it, because what they do all day, what they are allowed to do with their data, and what a mistake costs them are all different.

This section looks at AI through the lens of the work rather than the software. It explains what genuinely changes from one industry to the next, walks through how AI tends to show up in a range of common sectors, and gives you a way to work out which of it applies to you.

What you will find in this section

This is a hub page, not a ranking. Nothing here is presented as a best or worst industry for AI, because that question does not have a sensible answer: adoption depends far more on the specific task and the specific firm than on the sector label.

What you will find instead:

• The parts that are genuinely industry-specific, separated from the parts that only look that way.
• Short profiles of several common industries, covering the tasks where AI is most often applied and the work that tends to stay human.
• The constraints that matter most in regulated fields, described honestly rather than waved away.
• A method for applying this to your own industry, including ones not listed below.

If you are new to AI tools entirely, read getting started with AI first. It covers what these tools do, how to brief them, and what to avoid — and everything on this page assumes that groundwork.

What actually changes between industries

Four things vary by sector, and it is worth being precise about which is which, because they call for different responses.

The tasks worth doing. Every industry has its own repetitive, text-heavy work. An e-commerce business writes product descriptions; a consultancy writes proposals; a clinic writes referral letters. The underlying capability is identical — turning structured information into readable prose — but the task you would actually start with is not.

The cost of being wrong. A clumsy social post is embarrassing. An incorrect dosage, a misstated tax position, or a wrong measurement in a listing is a different category of problem. The higher that cost, the more review a draft needs before it reaches anyone, and the less suitable the task is for automation.

What you may do with the data. Confidentiality obligations, professional duties, and data protection rules are not uniform. A freelance copywriter and a financial adviser face very different questions before pasting a client document into a tool.

The software you need it to fit. Most industries run on sector-specific systems — practice management, property portals, accounting ledgers, learning platforms. Whether an AI tool connects to yours often matters more than how good its writing is.

Notice that only the first of these is about capability. The other three are about context, which is why a tool that transforms one business can be close to useless in another that does superficially similar work.

How AI shows up across common industries

The table below maps several industries to the kind of work AI is commonly applied to, and to the work that generally stays with a person. It is a starting point for your own thinking, not a prescription — the right answer for your firm depends on your clients, your obligations, and your existing systems.

Common AI applications by industry

Industry	Frequently applied to	Usually stays with a person
Marketing	Draft copy, campaign variations, briefs, repurposing one asset into several formats	Strategy, brand voice decisions, factual claims about the product
E-commerce	Product descriptions at volume, category copy, review summaries, listing translations	Specifications, pricing, compliance and safety claims
Finance and accounting	Transaction categorisation, document summaries, first-draft client explanations	Advice, filings, figures, and anything a regulator or client relies on
Healthcare	Administrative drafting, scheduling and correspondence, note structuring where permitted	Clinical judgement, diagnosis, and any patient-facing decision
Education	Lesson and resource drafting, differentiating material, feedback first drafts	Assessment decisions, safeguarding, and judging a student's actual understanding
Real estate	Listing descriptions, enquiry replies, area summaries, viewing follow-ups	Valuations, measurements, legal statements, and negotiation
Customer support	Drafting replies from existing help content, tagging and routing, summarising tickets	Complaints, refunds, exceptions, and anything a customer is angry about
Professional services	Proposals, client updates, meeting notes, turning research into readable summaries	Advice, scope and pricing commitments, and the judgement clients are paying for

Marketing

Marketing was one of the earliest places AI writing tools took hold, largely because the work is high-volume, text-based, and judged quickly. Drafting ad variations, outlining content, adapting one campaign message for several channels, and getting past a blank page are all tasks where a mediocre first draft still saves time.

The limits show up in two places. Output drifts toward generic unless you give it strong examples of how you actually sound, and anything it states about your product needs checking against reality rather than against what sounds persuasive. Our guide to AI tools for marketing works through the use cases in detail, and the AI writing tools category covers how to evaluate tools for this kind of work.

E-commerce

E-commerce has an unusual shape: the same writing task repeated hundreds or thousands of times. That makes it one of the clearer cases for both AI drafting and automation, because the saving scales with catalogue size rather than being a few minutes here and there.

Common applications include generating product descriptions from specification data, writing category and collection copy, summarising review themes, and adapting listings for different marketplaces or languages. The risk scales too: an error in a template repeats everywhere, and marketplace rules about accuracy and prohibited claims apply regardless of how the text was produced. Specifications, safety information, and pricing should come from your source data, not from a model’s guess. AI tools for e-commerce covers each of these workflows and the checks they need.

Finance and accounting

Finance splits cleanly between bookkeeping-style work and advisory work. On the bookkeeping side, pattern-matching tasks such as categorising transactions, flagging anomalies, and chasing invoices are well established, and many accounting platforms now ship these features directly. On the advisory side, AI is mostly useful for drafting — turning a technical position into a plain-English explanation a client can read, or summarising a long document before you review it.

What does not move is responsibility. Language models can state incorrect figures and non-existent rules with complete confidence, and a filing or a piece of advice carries professional and regulatory weight. The practical pattern is draft with AI, verify against the source, and have a qualified person own the outcome. AI business and productivity tools covers the admin side of this in more detail.

Healthcare

Healthcare is the clearest example of a sector where the constraint is governance rather than capability. The administrative burden is real — correspondence, scheduling, documentation, referrals — and that is where most realistic near-term use sits.

Anything touching clinical judgement or patient data sits behind rules that vary by country, employer, and role, and often behind approval processes and specific procured systems. If you work in a clinical setting, the first question is not which tool is best but what your organisation, regulator, and professional body permit, and whether the tool has been approved for that use. Personal experimentation with patient information is not a reasonable starting point.

Education

Teachers and trainers deal with a constant volume of material to produce: lesson resources, worksheets, explanations pitched at different levels, and feedback. AI drafting fits that shape well, particularly for producing several versions of the same explanation for students who need different entry points.

Two constraints are specific to the sector. Institutional policy on AI use — for staff and students — is often explicit and evolving, so it governs what you may do. And output about a subject needs checking by someone who knows it, because a confident, wrong explanation is worse than no explanation when the audience cannot yet tell the difference. Assessment decisions and anything involving student wellbeing stay with a person.

Real estate

Property work is a mix of high-volume writing and high-stakes accuracy sitting side by side. Listing descriptions, enquiry responses, viewing follow-ups, and neighbourhood summaries are all repetitive drafting tasks where a first draft helps.

The accuracy line is sharp, though. Measurements, tenure, legal status, and anything a buyer might rely on must come from verified records, and misdescription rules apply to marketing material in many jurisdictions. Treat AI as help with the prose around the facts, never as a source of the facts.

Customer support

Support is where the distinction between assisting and replacing a person matters most, because the customer is present. The dependable use is drafting: an AI tool proposes a reply based on your existing help content, and an agent approves or edits it. Summarising long ticket threads, tagging, and routing are similarly low-risk because a person still handles the response.

Fully automated responses are a bigger step and deserve deliberate scoping — what the bot may answer, how it hands over, and what happens when it does not know. The failure mode is confidently answering the wrong question to someone already frustrated. Customer-facing automation is covered in the AI automation and workflows category, including how to add review steps rather than removing them.

Small business and professional services

Consultants, agencies, trades, and other small firms share a pattern regardless of what they actually sell: one or two people carry sales, admin, delivery, and finance at once. The useful AI applications follow that shape — proposals and quotes, client updates, meeting notes, turning scattered research into something readable, and handling the routine correspondence that accumulates.

This group also has the least time to evaluate software, which makes it the most prone to accumulating subscriptions that never become habits. One tool used weekly is worth more than five trialled once. AI tools for small business works through these use cases job by job; best AI tools for small business covers specific options, and how much AI costs for a small business covers what the plans actually come to.

When industry context raises the stakes

Some sectors carry obligations that sit above any question of whether a tool works well. Regulated professions, anything involving health data, financial advice, legal work, and settings involving children all fall into this group.

In those contexts, a few things are worth treating as fixed:

• Permission comes before capability. What your regulator, employer, professional body, insurer, and client agreements allow determines the answer, not what the tool can do.
• A qualified person owns the output. AI assistance does not transfer responsibility for advice, a diagnosis, a filing, or a decision.
• Data handling is a documented decision. Check the vendor’s terms and your own obligations before sensitive information goes near a prompt, and record why the answer was acceptable.
• Approved systems usually beat better ones. A tool already procured and permitted is often more useful in practice than a more capable one you are not allowed to use.

None of this makes AI unusable in regulated work. It means the starting point is administrative and internal rather than client-facing.

Applying this to your own industry

If your sector is not profiled above, the method transfers. Work through it in this order:

1. List the repetitive text work in your week. Anything you write more than once a month in roughly the same shape is a candidate — quotes, updates, replies, summaries, listings, reports.
2. Sort by what a mistake costs. Start at the cheap end. You want to learn how the tool behaves where a bad result costs you nothing.
3. Check what you are allowed to do. Before real client or customer data is involved, confirm your obligations and the vendor’s terms. This is a five-minute check that prevents a serious problem.
4. Check what you already pay for. Sector software increasingly includes AI features on some plans. Upgrading a system you already use is usually simpler than adding another login.
5. Try one task for a fortnight. Judge it honestly: does editing the draft take less time and effort than writing it yourself? If not, stop using AI for that task rather than forcing it.
6. Only then consider automating. Automation multiplies whatever it is given, mistakes included. Do the task manually with AI’s help until it is boring, then look at AI automation and workflows.

The pattern that holds across every industry we have looked at is the same one that holds for individuals: use AI for the first draft, and keep the last decision human. What changes by sector is how much checking sits in between.

Where to go next

• New to all of this. Getting started with AI covers what these tools are, how to brief them properly, and a realistic first month.
• Your industry runs on writing. AI writing tools explains the categories and how to evaluate them.
• You want less admin. AI business and productivity tools covers meetings, scheduling, bookkeeping, and support.
• You are ready to remove repetitive work. AI automation and workflows is the hub for connecting tools together.

Guides in this section

The first industry guides go deeper into the workflows and constraints behind the profiles above:

• AI tools for marketing — planning, research, drafting, campaign variations, repurposing, customer insight, and reporting, with the line between assistance and review drawn explicitly.
• AI tools for e-commerce — product copy from your own data, support drafting, review summaries, merchandising assistance, and the accuracy rules that catalogue scale demands.
• AI tools for small business — customer communication, content, admin, research, sales support, and scheduling, organised by the jobs that fill a small business week.

More sector guides are planned. We would rather leave a gap here than fill it with something thin, and how we research and label what we publish is set out on our about and editorial process page.