Earlier business notes
Using AI without turning it into another obligation
How founders can choose a small AI experiment, evaluate its usefulness and keep responsibility for the result.
By Rajiv Sankarlall · Updated
If every conversation about AI makes you feel behind, pause before buying another tool. A useful experiment starts with a task you understand, a reason to change it and a way to tell whether the change helped.
This is a practical technology note. The coaching offered on this site is about founder burnout, values and purpose. It does not include building an AI employee or installing an operating system for your company.
Choose a task you can evaluate
Start with something small enough to check yourself: turning your own notes into a draft, organizing a set of questions or proposing a structure for a document. Avoid beginning with a task where you cannot recognize a convincing but incorrect answer.
Write down what good work looks like before you try the tool. For a draft, that might mean keeping the facts intact, sounding like your company and leaving you with less editing. Compare the finished work with your usual process, including the time spent checking it.
Keep the boundary visible
Decide what information you are comfortable sharing and check the provider's current terms and settings before adding confidential material. Start with invented or public data if the task allows it. Convenience is not a reason to skip that decision.
Also decide what the tool can suggest and what a person must approve. Generating an email draft and sending it to a customer are different responsibilities. Keep the approval step explicit rather than relying on everyone to remember it.
Keep or stop the experiment
After a few attempts, ask whether the output was useful, what mistakes appeared and whether the review burden was manageable. You do not need to keep a tool because it looked impressive in a demo. A disappointing experiment can still clarify where human knowledge matters.
For broader planning, NIST's AI Risk Management Framework offers guidance for considering risks throughout an AI system's use.
There is a personal question alongside the technical one: what do you want the saved attention for? If every efficiency becomes room for more work, your week may still feel the same. Make that choice deliberately. Values beyond the company is a place to begin.