You slap a 1-to-5 likelihood-times-impact sticker on every backlog item like that’s strategy, Residual Risk. A model can fill that grid and the compensating-controls bullets from the one-pager before standup. Then you still recite the leftover-risk line out loud, and the room hears last sprint’s same four numbers on the same sticker.
Your busywork’s shelf life~11 months
Estimated AI takeover of 3 exposed tasks. Whole-job replacement: uncertain.
Opens a text draft and downloads your roast image. Add it with LinkedIn’s photo button, then review and post. If the text is blank, paste the copied roast.
An illustrative midpoint of 3 medium/high-exposure task windows: now–6 months → 3, 6–18 → 12, 18–36 → 27. High exposure gets weight 2; medium gets 1. Low-exposure tasks are excluded. This gives ~11 months from the assessment date, not a countdown to losing your job.
Calculation: (3 × 2 + 12 × 1 + 27 × 1) ÷ 4, rounded to the nearest month. The task assumptions below must hold.
The whole role is unlikely to vanish because someone still owns tradeoffs, shipping risk, and what gets promised to buyers. Generated drafts can replace much of the paperwork, not the accountability.
Confidence: low. Timing is conditional; inferred duties may not match your actual workload.
Draft product requirements specs
Now–6 months
high exposure · Inferred from role
How AI takes over
AI models ingest tickets and notes, draft specs and acceptance criteria, shrinking blank-page writing and first-pass editing.
What has to happen first
Teams already paste transcripts into copilots; the gate is whether leaders accept first drafts without extra process.
Synthesize customer feedback themes
6–18 months
medium exposure · Inferred from role
How AI takes over
AI clusters interviews and support mail into themes and quotes, reducing hours spent tagging comments by hand.
What has to happen first
Needs clean access to call notes and CRM comments plus trust that summaries do not miss quiet but costly complaints.
Prioritize roadmap tradeoffs
18–36 months
medium exposure · Inferred from role
How AI takes over
Copilots score requests against goals and capacity, producing ranked lists so humans spend less time building the first ranking.
What has to happen first
Adoption waits on political buy-in, because ranking still needs a person who will own the no.
What still needs you
Someone still has to pick the painful no, stand behind a ship date, and take the blame when it slips.
Your next move
Run one cycle where you only edit an AI-first spec and spend the saved hours in live customer fights over scope.
The score averages three task ratings: low = 20, medium = 50, high = 80. It's an illustrative index, not a probability of losing your job.
Preview the downloadDownload the image and attach it to your own post.
