The case that it fades
The other side of the debate framed in Will Scrum survive the AI era?. This page argues the no deliberately and one-sidedly. The opposing view is the case that it survives.
Thesis: Scrum-as-practiced was calibrated to a world that AI is dismantling. Most of what teams actually do in the name of Scrum is overhead for constraints that no longer bind. Strip those away and what remains is too thin to keep the name.
Estimation and velocity measure the wrong organ
Story points proxied human effort, and velocity proxied a team's steady throughput of it. AI throughput is bursty and uncorrelated with how complex the work feels: a "5-point" task lands in twenty minutes; a "1-point" task fights you all day because the hard part was never the typing. Burndown charts then chart noise. Planning poker becomes a ritual that produces numbers no one should trust. An estimate of output is close to worthless when output is no longer the scarce thing.
The standup is theater when work is observable
The daily status sync existed because humans couldn't see each other's work in flight. When work streams through PRs and agent logs in real time, the verbal "what I did / will do / blockers" is recounting what is already visible. It survives as social ritual, not coordination — and a ritual defended on social grounds is a ritual that has lost its original job.
The sprint cadence is outrun
A two-week boundary was a sensible unit when an increment took weeks. When a feature is an afternoon, the sprint stops being a planning horizon and becomes an arbitrary administrative beat: re-planning ceremonies for work that already shipped, commitments that are stale by day three. The cadence the work actually wants is continuous, not fortnightly.
The roles collapse
Much of Scrum is scaffolding for coordinating a team — a cross-functional group of humans with a Scrum Master to grease the process and a Product Owner to arbitrate. One human directing agents does not need a standup with itself, a facilitator for a one-person ceremony, or a velocity to defend to a sibling team. At small scale the role distinctions dissolve into a single accountable human, and the framework built around those roles has little left to do.
It was always weak where AI presses hardest
Scrum is an execution framework: it assumes a groomed backlog already exists and optimizes the march through it. But when building is trivial, the scarce, decisive activity is discovery — what is even worth building, and is this the right thing? Scrum has almost nothing to say there. Ways of working organized around continuous discovery and validated learning answer the now-dominant question directly, and they may simply displace Scrum rather than wait for it to adapt.
Survival as a zombie is not survival
Scrum will persist in large organizations — but for reasons unrelated to whether it helps: auditability, contractor management, predictability for finance, certification economies, and something legible to sell executives. That is compliance ritual, not a living practice. Pointing at its persistence there and calling it "survival" confuses inertia with fitness.
What this side concedes
This is a steelman, not the whole truth. It concedes that something endures — transparency, inspection, intent, verification. But it argues that core predates Scrum, is not unique to it, and is better carried by lighter, verification-first practices. Keeping the Scrum label on it is nostalgia.
Bottom line: the parts of Scrum that teams spend the most time on are the parts AI makes obsolete, and the parts that endure don't need the brand. What replaces it will look less like a sprint and more like a verification gate.