Usually, yes. Most mainstream business software now either has AI built in or connects to AI through ready-made connectors and automation platforms such as Zapier and Make. The exceptions are older desktop systems and niche industry software without an API, a documented way for other software to exchange data with it. Those need exports, email workarounds or a vendor upgrade.
The practical job is an audit: half an hour with your list of subscriptions, working out for each tool which of four routes applies, roughly what it costs and what to ask the vendor. To make it concrete, there's an illustrative insurance broker with a typical mix of modern cloud tools and one stubborn old system.
Four ways AI reaches the software you already run
Route 1: AI built into the product. The simplest route, because there's nothing to connect. Copilot Chat is included with Microsoft 365 business plans, and Microsoft 365 Copilot Business adds AI across your own email and files for $21 per user a month on annual billing. Gemini is built into Google Workspace business plans. QuickBooks Online has AI agents for tasks such as categorising transactions and reconciliation; Xero's JAX assistant takes on jobs like bank reconciliation. HubSpot's Breeze AI runs on HubSpot Credits at $10 per 1,000. The limit: built-in AI mostly works inside that one product.
Route 2: connectors inside a chat assistant. A connector lets a chat assistant read from, and sometimes act in, another app. ChatGPT Business supports connected apps (OpenAI now calls them apps) including Google Drive, SharePoint, Slack, HubSpot and GitHub, and workspace admins decide which are switched on. Claude's Team plan includes connectors for Google Drive, Gmail, Google Calendar, Microsoft 365, GitHub and Slack; on Team and Enterprise plans an Owner has to enable them before members can use them. Claude's connectors run on MCP, the Model Context Protocol, an open standard for plugging tools into AI assistants, and many software vendors now publish MCP connectors for their own products. How to connect ChatGPT or Claude to your business apps walks through the setup.
What this route changes, in practice: an account handler at a broker asks ChatGPT, with the SharePoint app switched on, "Find our current terms of business and list what changed from the previous version." An illustrative answer:
The most recent version appears to be "Terms of Business v5 DRAFT (partner comments).docx", last edited three weeks ago. Compared with v4, it adds a $75 cancellation fee, changes the complaints contact and removes the clause on paying by instalments.
Useful, and wrong in an instructive way. The file it found is a partner's unapproved draft, in a folder shared with the whole firm years ago. The assistant did exactly what it should with the access it had. The fix is on the permissions side: approved documents in one folder, drafts where the whole firm can't see them, and questions that name the folder ("in Client Documents, Approved").
Route 3: automation platforms. These watch for something to happen in one app (a new email, a form submission, a new invoice), pass it through an AI step to classify, summarise or draft, then do something in another app. Zapier's directory lists more than 9,000 apps and Make's over 3,000, as of September 2026. Zapier starts at $19.99 a month billed annually for 750 tasks, where a task is each successful action step; Make starts at about $9 a month for 5,000 credits. Whether you need a platform at all, or your apps already link directly, is covered in native integrations versus Zapier.
Before choosing a plan, do the task sum. Take an illustrative office-cleaning company whose website form feeds a Zap: the new form entry triggers it, an "AI by Zapier" step decides whether the enquiry is domestic or commercial and pulls out the floor area, then two actions create a deal in the CRM and post a message to the team's chat. The trigger costs nothing. Each action counts as a task, and Zapier's AI step uses 1, 3 or 5 tasks a run depending on the model tier you pick. At 120 enquiries a month with the 1-task tier, that's 3 tasks a run and 360 a month, comfortably inside the 750 on the entry plan. Choose the 5-task tier and it becomes 7 tasks a run and 840 a month, which no longer fits. For a job this simple, the cheapest tier is usually enough; test it on 20 past enquiries before deciding you need more.
Route 4: a custom connection through an API. A developer connects your system's API to an AI model and you pay for the build plus usage, priced per token. It's the most flexible route and the most expensive, and it only makes sense when nothing ready-made fits.
Audit your tool list in 30 minutes
Start from your card statement or accounts: every software subscription the business pays for. For each tool, check four things.
- Built-in AI? Look at the product's features or pricing page for AI features, and which plan they come on.
- On Zapier or Make? Search each platform's app directory for the product and note which triggers and actions it offers.
- Public API, on your plan? Search for the product name plus "API documentation". Then check the pricing page, because many vendors only include API access on higher tiers.
- A connector for your chat assistant? Check the connector or apps directory of the assistant you use.
TOOL AUDIT
Tool | What we use it for | Built-in AI? (plan) | On Zapier/Make? |
Public API? (our plan?) | Chat connector? | Best route | Next step
Example row:
Xero | invoices, bank feeds, commissions | Yes, JAX | Yes |
Yes | check | Route 1 first | Switch on and measure for a monthA four-person physiotherapy clinic's audit might come out like this (illustrative; each answer is something to confirm on the vendor's own pages):
Google Workspace | email, documents, rota | Yes, Gemini (in plan) |
n/a | n/a | n/a | Route 1 | Switch on Gemini in Gmail for reception
Booking system | appointments, reminders | Note summaries, top plan
only | Yes: new-booking trigger, no cancellation trigger |
Top plan only | None | Route 3 for reminders | Ask vendor about a
cancellation trigger
Paper intake form | new-patient details | No | No | No | No |
Workaround | Replace with an online form that emails receptionThe pattern is typical: one tool with good built-in AI, one where the useful piece (the API) sits on a higher plan, and one that isn't software at all. The paper form is often the real blocker, and replacing it costs almost nothing.
Most small firms find three groups: tools with good built-in AI (use it), tools with connectors or automation support (connect them when a use case justifies it), and one or two old systems with no way in (see the workarounds below).
Which route to try first
| If the task... | Try first | Why |
|---|---|---|
| Happens inside one product (drafting in Word, coding in your accounts software) | Route 1, built-in AI | No connection to build or maintain; often included in your current plan |
| Means asking questions across documents, email or a CRM | Route 2, a chat connector | The assistant reads from the source when asked, with no workflow to build |
| Moves information from one app to another every time something happens | Route 3, an automation platform | Runs without anyone opening a chat window |
| Runs at high volume with no ready-made connection, or needs your own rules | Route 4, a custom build | Only once the task has proved its value on a cheaper route |
The order matters because each route up the list costs more to set up and more to keep running. A firm that starts at route 4 usually pays a developer to build something a route 1 feature would have done.
An insurance broker's stack, tool by tool (illustrative)
| Tool | What it holds | Route | Watch |
|---|---|---|---|
| Microsoft 365 | Email, documents, Teams | 1: Copilot Chat for all, Copilot Business for heavy users | Tidy SharePoint permissions first |
| Broker management system (desktop) | Clients, policies, renewal dates | No API on current plan: workarounds | Ask the vendor about API access and AI plans |
| HubSpot | Prospects, newsletters | 1: Breeze; 2: the HubSpot app in ChatGPT | Credits are used up per action |
| Xero | Invoices, commission income | 1: JAX; 3: Zapier if needed | Approvals stay with a person |
| Phone system | Call recordings | Check for built-in transcription or a Zapier app | Tell callers they're recorded |
| Insurer portals | Quotes and policy documents | Usually manual | Check each insurer's terms before any screen automation |
The resulting plan is modest and cheap. Renewal letters and client emails start in Microsoft 365 with Copilot. The broker management system can't connect directly, but it can export a weekly renewals report, so that report goes into a SharePoint folder where Copilot can draft the reminder letters from it. The broker asks the system's vendor about API access before next year's renewal. Nothing new is bought except Copilot seats for the two account handlers who write most.
A month in, the broker checks two things. The Copilot usage report in the Microsoft 365 admin centre shows whether both handlers are actually using their seats; a seat with no activity in 28 days is $21 a month for nothing, and is better moved to whoever writes the most client emails. Then the handlers count what they care about: how many of that month's 60 renewal letters went out from a Copilot draft, and how many needed more than a light edit. If most drafts needed rewriting, the fix is usually a better example letter in the folder, not a different tool.
A small bookkeeping firm on Google Workspace and Xero comes out differently, and more easily. Gemini in Gmail drafts replies to client queries. JAX takes on bank reconciliation work inside Xero. A Zapier workflow turns client emails that contain receipts into tasks in the firm's practice-management tool, with an AI step that reads the email and fills in the client name and month. The only gap is the client portal, which has no API; it sends a notification email whenever a client uploads a document, so that email becomes the trigger. Three routes, one workaround, and no custom build.
The AI step in that workflow is small and specific. For an incoming email reading "Hi, attached are the fuel and parking receipts for March, plus the van repair from early April", an illustrative output:
client_name: [matched from the sender's address]
month: March
document_type: receipts
note: email also mentions an April van repair; check which month
needs_review: yesThe needs_review flag is the part that matters. A first version of a workflow like this often forces a single month into the field, so the April repair quietly lands in March's task. Telling the AI to set needs_review: yes whenever an email covers more than one month sends those few to a person and lets the rest flow straight through.
Where connections break down
- No API, or an API only on the top tier. The cost of the connection can be the price jump to the vendor's highest plan.
- Read-only access. Some APIs and connectors let AI read data but not write it back, so a person still copies the result into the system.
- The AI sees what the user sees. A connected assistant can open every folder the signed-in person can, including ones nobody remembers sharing. Cleaning up SharePoint permissions before turning on Copilot shows how to fix that first.
- Costs that grow with volume. Every successful action step on Zapier is a task; triggers and filter steps aren't. A busy workflow with several AI steps uses tasks quickly.
- Messy data. Duplicate client records, inconsistent names and blank fields confuse AI as much as people.
- Silent failures. A connection that breaks after a software update may simply stop working without telling anyone. Stopping automations breaking silently covers the alerts to set.
Workarounds when a tool has no way in
- Scheduled exports. Many older systems can export a report as a spreadsheet file on a schedule. Save it to a shared folder and let the AI work from the folder. For the broker above, that's a weekly "renewals due in 45 days" export with columns for client, policy type, renewal date and handler; Copilot drafts one letter per row, and the handler checks each premium against the insurer's figure before anything is sent.
- Email as the bridge. If the system sends notification emails (a renewal due, a new booking), an automation can read those emails and act on them.
- Ask the vendor. Many vendors have an API or AI features on their roadmap and prioritise by customer demand. The email below takes two minutes.
- Screen automation, carefully. Robotic process automation (software that clicks through screens as a person would) works on systems with no API, but it breaks when the screens change and may breach the software's terms. Treat it as a stopgap.
- Replace it at renewal. If a core system blocks every route, make AI connectivity a requirement in your next software decision. What an API is and why it matters when you buy software explains what to ask for.
Subject: Integrations and AI for [PRODUCT] before our renewal
We use [PRODUCT] on the [PLAN] plan and are reviewing how it fits with
our other tools. Could you tell us:
1. Do you offer a public API? Which plans include it, and is it
read-only or can it update records?
2. Are you listed on Zapier or Make? Which triggers and actions exist?
3. Do you offer AI features inside [PRODUCT], or plan to? On which plans,
and at what cost?
4. Can we schedule an automatic export of [REPORT] to a shared folder?
5. What do your terms allow if data from [PRODUCT] passes through an
AI service via your API?
Thanks,
[NAME]Vendors' answers are rarely a simple yes or no, so read them for what they let you do. A reply to that email from the broker's system vendor might say (illustrative): "Our REST API is available on the Enterprise tier. We're listed on Zapier with triggers for New Client and New Policy; there are currently no actions. AI summaries are in beta for Enterprise customers." Translated: on the current plan there's no API; Zapier can react when a client or policy is added but can't write anything back; and the AI feature is a beta on a tier the broker doesn't pay for. So the scheduled export stays the main route, a Zapier trigger on New Policy could start a welcome email, and the Enterprise price goes into next year's renewal comparison rather than this year's budget.
Check permissions and data before switching anything on
Connecting AI to your tools is mostly a question of access, and access is where the risk sits. Before you enable a connector or build an automation, confirm three things: that an admin, not each member of staff, controls which connectors are on; that the people using it only have access to the folders and records they should; and that the tool's business terms say your data isn't used for training. Start with one team and read-only access where you can, then widen it. The AI security checklist for connecting email and files has the full list.
The short version of the whole question: the tools you already use are probably more ready for AI than you think, and the first step is rarely buying something new. It's switching on features your plans already include, and connecting one thing to another only when a real task needs it.
Questions about connecting AI to your software
Do I need to switch to Microsoft or Google to use AI?
No. Chat assistants such as ChatGPT and Claude connect to both, and automation platforms link thousands of other apps. If you already run on Microsoft 365 or Google Workspace, though, start there: Copilot Chat and Gemini are included in their business plans, so the built-in route costs little or nothing extra and needs the least setup.
Is it safe to connect AI to our email and files?
On business plans it can be, with two conditions. First, an admin should enable connectors deliberately rather than letting staff add them freely. Second, your file permissions must be tidy, because a connected AI can see everything the signed-in person can see. Start with one team, read-only access where possible, and a check of who can open which folders.
What does it cost to connect two apps with AI in between?
With an automation platform, often tens of dollars a month: Zapier starts at $19.99 a month billed annually for 750 tasks, and Make from about $9 a month for 5,000 credits, plus any AI usage. A custom connection built by a developer is a one-off project cost plus usage, and only worth it when no ready-made route exists.
Further reads
- What Connecting Two Business Apps Really Costs: Four Options — Four ways to connect two apps, costed side by side.
- ChatGPT Connectors: What They Can See in Your Drive and Inbox — Exactly what ChatGPT's connected apps can see in your files and email.
- Zapier vs Make vs n8n for AI Automation: Which Fits Your Business? — Choose between the automation platforms.
- What Does It Cost to Integrate AI Into Your Existing Software? — What a custom AI integration into your software costs.
- What Is a Webhook? Why Some Automations Run Instantly — Why some connections run instantly and others on a timer.
- How to Use Xero's JAX for Invoices and Cash-Flow Questions — An example of AI already built into accounting software.
- Do I Need AI in My Business? A Decision Guide for Owners — A decision guide for owners: the three situations where AI is worth it, when waiting is sensible, a ten-question score and a 30-day test.
- AI for Small Business Owners: A Plain-English Beginner's Guide — What an owner should know before starting with AI: how chat tools really work, the four kinds of product, data rules, costs and a first fortnight.
- What AI Implementation Actually Involves for a Small Business — The six pieces of work behind a small-business AI implementation, with rough hours for each and one workflow followed from baseline to handover.
- Barriers to AI Adoption in Small Businesses and How to Clear Them — Eight practical barriers that keep small firms from adopting AI, the cheapest way past each, and the few barriers you should respect rather than clear.
- Use the AI Already in Your Practice Software Before Buying More — Five stages to find and test the AI in your practice management, accounting and office software, then buy only for the gaps it leaves.
- AI Features Already in Your Software: What to Switch On First — Thirteen built-in AI features in Microsoft 365, Google Workspace, Xero, QuickBooks, Canva, HubSpot and Shopify, ranked by when to switch each one on.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Anthropic's Claude support articles on connectors and the Team plan; OpenAI's help articles on connected apps and SharePoint; Zapier app directory; Make integrations page; Microsoft 365 Copilot, Google Workspace, Zapier, Make and HubSpot pricing pages; Xero JAX and QuickBooks Online AI agent pages.