Building products with AI
Keep control of your AI product when you work with a builder
Three things you must own when you hire someone to build an AI tool: the code, the data, and the accounts that run it.
· 5 min read · Product strategy · Vendor relationships · Data ownership
When you hire a team to build an AI product for your business, you're making a bet. You're betting they'll understand what you need, build it right, and stay available to help you later. But what happens if they don't, or if the relationship ends.
The smart move is to own three things from day one: your code, your data, and your accounts. This isn't paranoia. It's the difference between owning a tool and renting one.
Own the code
The code is the recipe. It's the instructions that make your AI tool work. If you don't own it, you're dependent on the builder to make any change, no matter how small.
A dental clinic in Kelowna hired a developer to build a tool that reads patient forms and pulls out key details. Six months in, the clinic wanted to add a new field. The developer quoted $3,000 and a two-week wait. If the clinic had owned the code, they could have hired someone else for $500 and gotten it done in two days.
Owning the code means:
- You get the full source code, not just the running tool.
- You can read it and understand how it works.
- You can hire someone else to change it if the original builder is too slow, too expensive, or unavailable.
- You're not locked into one person or one team.
Put this in your contract: "The client owns all source code created during this project and receives a complete copy at delivery."
Own your data
Your data is your business. It's your customer information, your transaction history, your documents, your feedback. If the builder keeps your data on their servers, you have a problem.
A property manager in Surrey built an AI tool with a vendor that learned from past leases to write new ones faster. After two years, the manager wanted to switch builders. The vendor said: "Your training data lives on our servers. You can't have it."
The manager lost two years of work. They had to start over with a new builder.
Owning your data means:
- Your data stays in your own cloud account (AWS, Google Cloud, or similar) or on your own servers.
- The builder can access it to do the work, but you control it.
- You can export it all, in a standard format, whenever you want.
- If the relationship ends, you keep everything.
Put this in your contract: "All data generated or collected by the tool remains the property of the client. The client retains the right to export all data in standard formats at any time."
Own your accounts
Your accounts are the keys to the kingdom. If the builder controls the cloud account, the API keys, or the login credentials, they control your tool.
A plumber in Richmond hired a team to build an AI scheduling tool. The team set it up on their own AWS account. When the plumber wanted to hire a new IT person to help manage it, the original team said no: "You can't have the account credentials."
The plumber was locked out of their own tool.
Owning your accounts means:
- You own the cloud account (AWS, Google Cloud) where the tool runs.
- You own the API keys and secrets that make the tool work.
- You own the login credentials for any service the tool uses.
- The builder can have temporary access to do the work, but you control the master keys.
Put this in your contract: "The client owns all cloud accounts, API keys, and credentials. The builder will use accounts provided by the client or will transfer full ownership to the client upon project completion."
What this looks like in practice
A 12-person software company in Gastown hired a builder to create an AI feature that summarizes customer support tickets. Here's what they did right:
- Code: The builder delivered the full source code. The company stored it in their own GitHub account.
- Data: The builder accessed the company's own AWS account. The support tickets stayed in the company's database, never copied to the builder's servers.
- Accounts: The company created temporary AWS credentials for the builder. When the project ended, the company revoked those credentials and kept the master keys.
Six months later, the company wanted to add a new feature. They hired a different builder, gave them access to the code and the same AWS account, and got it done in three weeks. No renegotiation. No lock-in.
The conversation to have
You don't need to be technical to ask these questions. You just need to ask them before the work starts.
Ask your builder:
- "Will I own the source code, and can I get a copy when we're done."
- "Where will my data live, and can I export it whenever I want."
- "Who will own the cloud account and API keys."
If the builder hesitates, that's a signal. A good builder will say yes to all three. They know you're being smart.
If they push back, ask why. Sometimes there's a real reason (they need to manage security, for example). But if they say "that's not how we work" or "it costs extra," keep looking.
Where to start
- Before you hire anyone, write down what you want the AI tool to do and what data it will use. Then send your shortlist of builders this question: "Can I own the code, my data, and the accounts that run this tool."
- Ask your builder for a sample contract or terms. Look for the three things mentioned above. If they're not there, ask them to add them.
- If you already have a builder working on something, schedule a 30-minute call and ask these three questions. Get the answers in writing via email.