batteriesincluded.com · Questions & Answers

We have 30 employees and every department wants AI. How do we choose the first project?

Choose one recurring workflow with a measurable burden and a willing internal owner, not the department with the loudest request. For a 30-person company, the first project should prove that a specific piece of work can improve without requiring a company-wide rollout.

Start by checking the AI capabilities already available in your existing software. Ask each department to describe one practical outcome, such as reducing the time spent preparing a recurring report or organizing incoming requests. These are candidate outcomes, not promises of results.

Compare the candidates against the same questions:
- Does the work happen often enough to observe improvement?
- Can we measure its current manual effort or cycle time?
- Are the necessary inputs available?
- Can someone check whether the output is correct?
- Will a named internal champion own the workflow after coaching?

Prefer the candidate with the clearest answers and the fewest dependencies. Defer projects that require several departments to change their processes before anyone receives value.

For an EOS-run company, turn the selected improvement into a quarterly Rock with one owner and a clear definition of done. Other leaders can identify opportunities without launching competing builds.

Use Lovable or Replit when a custom application is justified by the workflow. Choosing the building platform comes after choosing the business outcome. The first win should be a better operating routine, not simply a new application.

Category: First AI Project

← All questions