Why a 90-day horizon works
Ninety days is not how long a career transition takes; full transitions commonly take far longer. Ninety days is the right length for one cycle of testing a direction. It is long enough to learn something real, produce a tangible artifact, and gather feedback, yet short enough that you committed to an experiment rather than a life change. If the direction proves wrong, you have lost one quarter, not two years of night courses toward a target you never validated.
The horizon also fixes the two standard failure modes of career plans. Plans with no deadline dissolve into someday. Plans framed as become a different person by June collapse under their own weight at the first busy week. A 90-day experiment has a defined end, a defined output, and a scheduled decision: continue, adjust, or stop. That decision point is the real product. You finish knowing something about the path that you could not have known by reading about it.
How to pick one target experiment
Start from evidence, not aspiration. If you have mapped your tasks' AI exposure, your map already shows which zones deserve investment: skills that serve your field's judgment, review, and relationship work. From those, choose exactly one target for the quarter. The selection test has three parts: it connects to work you already know, you can produce a demonstrable artifact within 90 days, and someone real, a hiring manager or your own boss, would care about the result.
One target means one. Splitting a quarter across three skills produces three fragments no employer can evaluate. Write the experiment as a falsifiable sentence: by day 90 I will have produced X, shown it to Y, and I will continue only if Z. A vague version like get better at data cannot fail, and what cannot fail cannot teach. Expect the first week's version of this sentence to be wrong; revising it once real contact begins is normal, not a setback.
Collecting evidence employers accept
Employers discount claims and certificates of attendance; they weigh work they can inspect and outcomes someone else vouches for. Strong evidence from a 90-day cycle looks like: an artifact solving a real problem, ideally from your actual workplace; a before-and-after record showing what changed because of your work; and a person willing to confirm it happened. A workflow you redesigned, with its documented playbook and logged results, outweighs a stack of course completions because it survives questioning in an interview.
Build the evidence as you go rather than reconstructing it later. Keep a dated log of what you did and what resulted; screenshots and documents beat memory. Ask for feedback in writing while the work is fresh, since a two-line email from someone who used your work is checkable in a way a memory is not. At day 90, assemble everything into a single short narrative: the problem, what you did, what changed, what you would do next. That narrative is simultaneously your review document and your interview answer.
