batteriesincluded.com · Questions & Answers

We are heading into our busiest season. Should our 30-person company start its first AI project now or wait?

Start now only if you can run a narrow pilot without making critical operations depend on it. If employees cannot spare time to review results, postpone live deployment and use the period to document the workflow and its current effort.

Peak demand can expose repetitive work clearly, but it also raises the cost of mistakes and leaves less room for coaching. Do not replace a working process simply because an AI prototype is quick to build.

Choose a recurring task whose output can be checked before it affects a customer, payment, or operational commitment. Run the prototype alongside the existing process on a limited sample. This gives you evidence without forcing a company-wide change during a demanding period.

Set boundaries before work begins:

- Name the internal champion and confirm actual availability.
- Define the manual effort or cycle time you want to reduce.
- Limit the pilot to a specific business request.
- Cap usage costs and the time employees spend testing.
- Keep a clear route back to the existing process.
- Stop testing if it interferes with service or essential work.

In quarterly planning, make the implementation outcome match available capacity. A Rock might reasonably finish with a validated pilot and a rollout decision rather than full deployment.

The recommendation is not to wait for a perfectly quiet quarter. It is to separate learning from operational dependence. Begin with useful capabilities already available in your software where possible, prove a small improvement, and expand when the team can support the change.

Category: First AI Project

← All questions