What lock-in really costs when you adopt AI

Lock-in is what happens when a vendor makes it too expensive or too difficult to leave. It's not always intentional. Sometimes it's just how the tool is built. But the result is the same: you're stuck.

Here's what lock-in looks like in practice.

A property manager in Surrey signs up for a vendor's AI system to organize tenant documents and generate lease summaries. The system works well for six months. Then the vendor raises prices by 40 percent. The manager wants to switch to something cheaper. But the vendor won't export the data. The code that runs the summaries is locked inside the vendor's account. The manager has two choices: pay the new price or rebuild the whole thing from scratch. That's lock-in.

Or a dental clinic in Kelowna builds a chatbot with a vendor to handle appointment bookings. The vendor owns the code. The clinic's patient data lives in the vendor's database. When the vendor gets bought by a larger company and the product shuts down, the clinic has no way to move the chatbot to a new platform. They have to rebuild it. That's lock-in.

Or an accounting firm in Burnaby uses a vendor's AI tool to extract data from client invoices. The extracted data lives only in the vendor's system. The firm can't pull it into their own accounting software without manual work. They're dependent on the vendor's export feature working the way they need it to. That's lock-in.

Why it costs more than you think

The monthly fee is not the cost of lock-in. The cost is what happens when you want to leave.

If your data is trapped, you pay to migrate it—or you lose it and start over. If the code is locked, you pay a developer to rebuild what you already have. If your accounts are controlled by the vendor, you lose access to your own work and your own customers' information. A small business might spend $5,000 to $20,000 to escape lock-in. A larger firm might spend $50,000 or more.

But the real cost is often invisible. You stay with a tool that no longer fits because leaving costs too much. You can't negotiate with the vendor because you have nowhere else to go. You miss the chance to use something better because you're locked into something adequate. That's the cost of lock-in: lost options.

For established businesses, that's a serious problem. You've built your operations carefully. You know what works. You can't afford to be forced into a tool that doesn't fit just because leaving it is too hard.

The three things to own

You can avoid lock-in by making sure you own three things from the start.

Your data stays in your systems. The AI tool should read your data where it already lives—your email, your file server, your CRM, your accounting software. When the tool generates new data (summaries, predictions, extracted information), that data should live in your systems too, not locked in the vendor's database. You should be able to pull it out anytime in a standard format. A plumber in Richmond should be able to export job quotes that the AI tool generates and use them in his own quoting software. A distributor in Calgary should be able to take extracted inventory data and put it straight into their system.

Your code lives where you can see it. If a developer builds custom code for you, you should own it. That means the code lives in a repository you control (or can access), not hidden inside a vendor's platform. You should be able to read it, modify it, or move it to a different developer if you need to. A 12-person software company in Gastown should never build a tool where the source code is locked away. You need to be able to understand what you've paid for and change it later.

Your accounts belong to you. You should own the login credentials and the administrative access to any tool that touches your business. You should be able to invite team members, remove them, export your work, and cancel without asking the vendor for permission. If the vendor goes out of business or changes their terms, you don't lose access to your own data or your own work.

How to ask the right questions

Before you sign a contract or hand over data, ask these three questions in writing and get clear answers.

• Where does my data live, and can I export it anytime in a standard format.
• Who owns the code that's being built for me, and can I access it or move it later.
• Who controls the login credentials and administrative access to this tool.

If the vendor hesitates, that's a signal. If they say the code is proprietary and you can't see it, that's a signal. If they say the data lives in their system and you can't export it, that's a signal. Those aren't deal-breakers—sometimes the tool is worth the risk. But you should know what you're giving up.

For most established businesses, the answer is no. You can't afford to lose control. You have too much at stake.

Where to start

• Before your next AI project, write down the three things you need to own: your data, your code, and your accounts. Use that as a checklist when you're evaluating vendors or developers.

• Ask one vendor or developer you're considering to answer the three questions in writing. See how they respond. If they're evasive, move on.

• Talk to one peer in your industry who's already adopted an AI tool. Ask them what they wish they'd known about data ownership and lock-in before they started.