How-To Guide

The AI You're Already Paying For: How to Find AI Features Hidden in Software You Own

Your office suite, accounting package, CRM and design tool have all shipped AI features — here is how to find them, switch them on, and only then decide what to buy.

B Biztrategy Published 27 October 2026 · 8 min read
Flat lay of a workspace with laptop, documents, and cardboard boxes for e-commerce operations.

Here is a pattern worth noticing. A small business decides to get serious about AI, and the first move is almost always to buy something: a new assistant, a new writing tool, a new meeting notetaker. Six months later the picture looks like one owner put it — "six AI subscriptions, no coherent workflow" — and the software bill has grown by a few hundred euros a month while the actual work has barely changed.

What gets missed in that rush is that most of the AI a 5-to-50-person business needs in its first year is already bundled into software it pays for right now. Your office suite, your accounting package, your CRM, your helpdesk, your design tool and your e-commerce platform have all shipped AI features over the past two years — usually inside plans you already hold, or as a modest add-on to them. This guide is about finding that AI, switching it on, and only then deciding whether you genuinely need to buy anything else.

Why small businesses buy AI they already own

This is not carelessness. Three things make bundled AI almost invisible to the people paying for it.

Vendors market the upgrade, not the inclusion. Your software providers have a strong commercial interest in telling you about the premium AI tier. They have very little interest in emailing you to say "the plan you are already on now does transaction categorisation automatically." That announcement lands in a release note nobody reads.

AI features arrive silently. A sparkle icon appears in a toolbar, a "summarise" option turns up in a right-click menu, a "draft a reply" button appears under a support ticket. If you were not looking on the day it shipped, you will not notice it for months.

The noise is all on the new-tool side. As one owner put it: "there's a new AI tool every week." Almost none of that noise is about the capability sitting dormant in software you already renewed.

The cost is not abstract. A dedicated AI assistant at roughly €20–30 per user per month across eight people is €1,900–2,900 a year. If three of those seats do a job your accounting software and office suite already do, that is real money spent twice on one capability. Our breakdown of what AI actually costs a small business has the full budget picture.

The AI already sitting in your stack

Feature names and plan packaging change constantly, so treat this as a map of what to look for rather than a product catalogue — then verify what is actually available inside your own admin panel.

Office suite, documents and email

Drafting and rewriting in documents, summarising a long email thread, turning a transcript into action points, generating a spreadsheet formula, drafting a slide deck. For most teams this is the densest pocket of unused AI in the business, because almost nobody has explored their office suite's AI menu.

Accounting, invoicing and payments

Automatic transaction categorisation, data extraction from photographed receipts and supplier invoices, duplicate and anomaly flagging, cash-flow projection, drafted chasing emails for overdue invoices. Finance tools have had machine learning inside them for years; what is new is that it is addressable in plain language.

CRM and sales tools

Email and follow-up drafting against a contact record, call summarisation, pipeline roll-ups in prose, next-best-action suggestions, deduplication and record enrichment. The advantage is that the AI can see your actual deal history, which no external chatbot can.

Helpdesk and support

Suggested replies drawn from your own past tickets, summarisation for handover, automatic tagging and sentiment flags, help-centre articles drafted from resolved tickets. If you are considering a support chatbot, check what your helpdesk already includes before you shop. E-commerce, booking and project tools follow the same pattern: product description drafting, review summarisation, note-to-task conversion, status roll-ups.

Design, marketing and social

Background removal and object erasing, automatic resizing of one asset into every channel format, headline suggestions, image generation, ad and caption variants. This is often the biggest surprise: the capability they were about to buy is a button in software they already have.

A fast way to build the shortlist: paste your tool names into an AI assistant with this prompt — "For each of these business tools, list the AI features it has shipped and where in the interface they usually sit. Say whether each is typically included in standard plans or sold as an add-on, and flag anything you are not confident about." Treat the output as a to-check list, not as fact: packaging changes constantly, and you are confirming each item in the admin panel anyway.

The 90-minute bundled-AI audit

Every guide tells you to "use AI" without saying how, so here is the sequence. Block out 90 minutes and work through it once.

  1. List what you already pay for (15 minutes). Pull three months of card and bank statements and write down every piece of software over €5 a month, including anything one person expensed. The list is usually longer than the owner expects.
  2. Open each tool's settings and search "AI" (30 minutes). Then scan the vendor's "what's new" page for the last six months. Two minutes per tool is enough — you are looking for a feature, not an opinion about it.
  3. Write three columns, not a report. Tool / AI feature / the job in our week it could do. If you cannot name a real recurring job in the third column, the feature does not go on the list — interesting is not the same as useful.
  4. Note tier and price. Mark each feature as included on your plan, an add-on at a stated price, or an upgrade. You now know what you own versus what you would be buying.
  5. Check the data terms (5 minutes per tool). Whether your content trains models, where processing happens, which admin toggles control it. Settle this before the feature touches client data — our AI and GDPR guide for small businesses covers what to look for.
  6. Pilot exactly two features for two weeks. One internal and low-risk, such as receipt extraction. One customer-adjacent but human-reviewed, such as drafted support replies. Two is the limit on purpose; five pilots means no pilot.
  7. Write down the result and act on it. Minutes saved per week, quality good enough or not, keep or switch off. A feature that saved nothing gets turned off and crossed out, which is a successful outcome, not a failed one.

If the audit also turns up standalone AI subscriptions, run the companion exercise in how to audit your AI tool stack — that one rationalises the AI you bought, this one inventories the AI you own without knowing it.

Built-in or buy something new: three questions

Once the list exists, every row needs a decision. Three questions settle almost all of them.

1. Is the work already happening inside that tool? Bundled AI wins on data gravity. The AI inside your CRM sees every previous conversation with that client; a general assistant sees only what you paste in. For jobs that live in one system — categorising transactions, summarising a ticket, drafting from a contact record — built-in beats better-but-elsewhere, because the alternative means copying data out and the answer back.

2. Would a dedicated tool be ten times better, or ten per cent better? Ten per cent is not worth a new subscription, a new login, a new vendor review and a new thing to train people on. Ten times usually is. Insist on seeing that difference in a two-week trial rather than in a feature table.

3. Who will own it after the trial? A feature nobody owns is unused within a month. One named person per feature, responsible for whether it stays — our piece on who should own AI in a small business covers assigning that without creating a committee.

Where bundled AI genuinely falls short

None of this makes built-in features sufficient on their own. They have real limits, and ignoring them is how businesses end up with a stack that cannot do the interesting work.

  • Depth and length. A sidebar assistant is not built for reasoning across a 40-page contract or a messy data export. A good general-purpose assistant still wins clearly on hard thinking and long inputs.
  • Anything that crosses tools. No bundled feature reads your CRM, your accounting ledger and your inbox at once. Cross-system work needs either a general assistant you feed deliberately or a proper integration — see how to connect AI to your business tools.
  • Prompt control. Most built-in features expose a button, not a prompt — you get the vendor's idea of a good summary, with little room to shape it.
  • Brand voice. Bundled drafting tends to produce competent, generic prose. Getting output that sounds like your business takes the kind of iteration covered in making AI writing sound like your brand.

So the practical answer for most small businesses is a two-layer stack: every bundled feature that earns its place switched on and owned, plus one good general-purpose assistant on a paid plan for the thinking work. That is a smaller bill and a more coherent workflow than five specialist subscriptions. For which layer deserves the money, free versus paid AI tools has the break-even maths.

A worked example: a nine-person consultancy

An illustrative composite, but a shape we see repeatedly. A nine-person consultancy was spending roughly €210 a month across five AI subscriptions, added one at a time by three different people — "I subscribed to seven different AI tools over six months," as the owner put it.

The audit took an afternoon and turned up four bundled features already on paid plans: receipt extraction and transaction categorisation in the accounting package, meeting summaries and drafting in the office suite, suggested replies in the shared inbox, asset resizing in the design tool. Three of the five standalone subscriptions did those exact jobs less well, because they sat outside the systems holding the data.

They cancelled those three — about €125 a month, roughly €1,500 a year — kept one general assistant on a team plan for proposal and analysis work, and spent the recovered budget on training people to use what remained. The real win was not the saving: it was that the stack finally fitted in one sentence, so new staff could be told what to use for what. If training is the gap in your setup, how to train your team to use AI is the next thing to read.

The bottom line

Before buying another AI subscription, spend 90 minutes finding out what you have already bought. List your software, search each admin panel for AI, keep only the features that map to a real recurring job, check the data terms, pilot two, measure. Then buy — deliberately, once, for the jobs bundled features genuinely cannot do.

The businesses getting real value from AI are rarely the ones with the most tools. They are the ones who can say in a sentence what each tool is for — and who stopped paying twice for the same capability.

Where does your business stand on AI?

Take the free 3-minute AI Readiness Quiz and get a personalised score with your next steps.

Take the Free Quiz →