Before you add another AI subscription, check the ones you have. Most business software added AI features over the past two years, and many firms pay for them without knowing they exist.
Key Takeaways
- Most mainstream business software now ships with some AI features, such as drafting, summarizing, categorizing and search, and some are included in the plan you already pay for.
- Check what you own before buying anything new, because a built-in feature costs nothing extra to connect and already sits inside your data.
- Spend one hour reviewing the release notes and settings of your 5 most-used tools and write down what each offers.
- Switch on features that save time on work you do weekly, and leave off any that need new data permissions until you have checked them.
- As of October 2026, vendors often put their strongest AI features on higher plans or sell them as add-ons, so confirm the cost of each before you rely on it.
- Buy a separate tool only when a built-in feature fails your test on a real task.
Why to Look Inside Your Software First
A built-in feature has three advantages over a new tool. It already has access to the data it needs, so there is nothing to connect. It runs under permissions you have already set. And it adds no new vendor to review, no new login and no new bill.
It also has a weakness. Built-in AI tends to be narrow, tuned for the one job the software does. It will draft a reply in your help desk and know nothing about your accounting.
For a founder with no operations leader, narrow and already connected usually beats broad and needing setup. The guide to a lean AI tech stack shows why fewer tools with clear jobs hold up better.
Where AI Features Hide in Common Business Software
The same handful of features appears across most categories. Look for these.
| Software type | AI features to look for |
|---|---|
| Email and calendar | Draft replies, summarize threads, suggest meeting times |
| Accounting | Transaction categorizing, receipt reading, anomaly flags, report summaries |
| CRM | Call and email summaries, next-step suggestions, record clean-up, lead scoring |
| Help desk | Suggested replies, ticket summaries, tagging and routing |
| Project management | Task summaries, status updates, drafting from notes |
| Document and file tools | Summaries, search by meaning, drafting from templates |
| Meeting and video tools | Transcripts, summaries, action item lists |
| Marketing and website tools | Draft copy, subject line suggestions, image generation |
Not every product has all of these, and names vary. Features also come and go, so treat the table as a checklist for your own review.
A One-Hour Review of Your Five Most-Used Tools
Do this once, then repeat every six months.
- List your five most-used tools. Use the ones your team opens daily, not the full subscription list.
- Search each vendor’s release notes or help center for “AI”. Most vendors keep a page for new features. Search the tool’s name plus “AI features”.
- Check your plan. Note for each feature whether it is included, limited, or sold as an add-on.
- Note what data it uses. Some features read only the record in front of you. Others read your whole account.
- Write one line per feature. Record the name, the cost, the data it touches and the job it could help with.
Put the result in a shared sheet. It becomes the starting point for an AI tech stack audit, and it shows which tools overlap.
Which Features to Switch On First
Rank features by how often you do the work and how cheap a mistake is. Good first choices are:
- Summaries of meetings and long threads. The risk is low and a person can check the summary against the source.
- Drafting replies in your help desk or inbox. A person reads and sends each one.
- Categorizing transactions in accounting. The bookkeeper reviews the suggestions, and a wrong category is easy to fix.
- Cleaning up CRM records. Merging duplicates and filling blank fields saves real time, as long as you review a sample.
Leave off anything that sends messages or changes records without a person seeing them first. The guardrails guide explains why limits on what a tool may change should come before you turn on its autonomy.
What to Check Before You Switch One On
Features that look harmless can change who sees your data. Ask four questions for each.
What data does it read? A feature that reads your whole mailbox is a bigger decision than one that summarizes a single thread.
Where does the data go? Read the vendor’s terms on whether inputs are used to train models and how long they are kept.
Who can use it? Many features respect existing permissions. If your permissions are loose, the feature will expose that.
What does your client contract say? Some contracts restrict how client data is processed. Check before enabling anything that reads client records.
The AI data security guide goes through these questions in more detail, and the acceptable use policy template gives your team a written rule for what goes where.
A Worked Example: A 10-Person Service Firm
Take a 10-person consulting firm with email, accounting, a CRM, a project tool and a video meeting service. The founder runs the one-hour review and finds four things. The video tool already produces meeting summaries on the current plan. The accounting software categorizes transactions but charges extra for report summaries. The CRM offers call summaries only on the next plan up. The email system drafts replies on a paid add-on.
The firm switches on the meeting summaries today, since they cost nothing extra and a person reads each one. It has the bookkeeper try the transaction categorizing for a month and check a sample of 50. It leaves the CRM and email add-ons off, because the combined cost per seat would exceed the time they save for a team this size. No new subscription was needed, and the founder has a one-page list to revisit in six months.
Your list will look different, and it will change as vendors release new features every few months. The shape of the exercise stays the same: find, cost, test, decide. The scenario above is illustrative.
How to Compare the Cost of Built-In and Separate Tools
Built-in does not always mean free. The cost often sits in a plan upgrade or an add-on per seat.
Compare like with like. For a built-in feature, take the extra plan or add-on cost per user per month and multiply by the users who will use it. For a separate tool, add its subscription, the time to connect it and the time to maintain the connection.
Then compare both with the hours saved. If a $20 per seat add-on saves one person two hours a month at $40 an hour, it pays for itself. If it saves 20 minutes, it does not. These figures are illustrative, so use your own. The free versus paid guide covers where the line usually falls.
When a Separate Tool Is Worth Buying
Test the built-in feature on three real tasks first. If it handles them well enough that a person only needs to review, use it.
Buy something separate when the built-in feature fails on a task you do often, when you need to work across several systems the vendor does not connect, or when the job needs more control than the feature gives. The off-the-shelf versus custom builds guide and the guide to evaluating AI tools help with that decision.
Each tool you add also adds to the sprawl described in the AI tool overload guide, so make the new one earn its place. For the wider picture, start with the pillar on choosing an AI tech stack.
Frequently Asked Questions
Does my accounting software already have AI?
Most mainstream accounting products now include some AI features, such as transaction categorizing, receipt reading or report summaries. Check the vendor’s release notes and your plan, since some features need a higher tier.
Are built-in AI features free?
Some are included in your current plan and some cost extra, either as a plan upgrade or a per-seat add-on. Confirm the cost on the vendor’s pricing page before relying on a feature.
Is it safe to turn on AI features in my CRM?
It can be, if you check what data the feature reads, where that data goes and what your client contracts allow. Start with features that only summarize or suggest, and review a sample of the output.
How do I find out what AI my software has?
Search each vendor’s help center or release notes for AI features, then check your plan. An hour on your five most-used tools covers most of it.
Should I still buy a separate AI tool?
Only when a built-in feature fails a real task you do often, or when the job spans several systems the built-in feature cannot reach. Test the built-in option first.
Book an AI readiness audit to review the tools you already own before you add another subscription.
