Answers

Our data sits across four systems and nobody agrees which one is right. Does that have to be fixed before AI is any use?

Zachary King, Cofounder and CEO at Works · LinkedIn

Short answer

No. You don't have to fix all your data first. You need to decide which system is allowed to be right. Set a rule for each record type, such as customer, job or project: one system is the source of truth and the others defer to it. That's a decision, not a project, and every automation after it gets smarter.

Do this week

Take the record type your team argues about most and put its owner system in writing.

Transcript

Our data sits across four systems and no one can agree which one is right. Does that need to be fixed before AI is of any use?

No, you don't have to fix everything first. You just need to decide which system is allowed to be right. Waiting for the perfect data layer, or layers, in your company is where AI projects go to die. What you actually need is a rule for each type of record: customer, job, project. One system is the source of truth and the others defer to it. That's a decision, not a project. Make that once and every automation after that gets smarter.

We run this inside our own business. The CRM is the truth for pipeline. The finance system rules for money. When two numbers disagree, nobody debates it, because we have a rule in place that the owner system is right. Every workflow that we build runs on that.

So this week, take your most argued record type and put its owner system in writing. That's it. System of record. That's the data cleanup that matters. Then you're off to the races.

Want this built inside your business?

Tell us where the work piles up. We will show you what it takes to ship it.