The quiet death of story points in the agentic era

Story points were a brilliant invention for the constraint of their era. The constraint was team capacity. A team had a fixed amount of human-hours per sprint. The job of the Product Owner was to decide which subset of work fit into that capacity. Story points gave teams a relative-estimation language that abstracted away from time and made the prioritisation conversation faster.

It worked. For 20 years, story points and velocity charts ran the modern Agile delivery process. The framework was right. The metric was right.

The constraint changed in 2024.

What changed

Agents are now competent enough to do meaningful work. Not all the work. Not perfectly. But enough that team capacity is no longer the binding constraint for most knowledge-work delivery.

If you have an engineering team of 8 humans and you add agents that can draft code, write tests, refine backlog items, summarise retros, and propose architecture trade-offs, your effective capacity for output is no longer 8 humans times sprint length. It is closer to 8 humans times sprint length times some multiplier between 2 and 5 depending on how mature your agentic workflow is.

Capacity is now elastic. Velocity is now meaningless as a comparison metric, because two teams with the same human velocity can be operating at 2x throughput differences when you factor in agent contribution.

The new binding constraint is clarity of intent. Teams that can clearly state what they want shipped, with what trade-offs, against which success criteria, get the agents to do the work. Teams that cannot drown in agent output that does not solve the right problem.

The fastest team is no longer the team with the highest velocity. It is the team with the clearest intent.

The new unit of value

If velocity is dead, what replaces it? In the AI-native Agile operating model we teach at Agile Visa, the new unit of value is Concept to Cash time: the elapsed time from a customer-validated concept to recognised revenue.

Concept to Cash measures the right thing for the new constraint. It rewards clarity of intent because clarity compresses the cycle. It rewards genuine customer validation because it counts time from validated concept. It rewards delivery to revenue because building features that customers do not buy adds no value to the metric.

It also strips away the easy gaming that story points enabled. You cannot inflate Concept to Cash by counting more tickets. You cannot demonstrate progress by closing trivial work. You either compressed the cycle or you did not.

What this means for the Product Owner role

The Product Owner role rewrites itself first. In the old model the PO's job was to write good stories and slice them small enough that the team could deliver them within a sprint. Both of those tasks are now agent-assisted.

The PO's job in the AI-native model is to be ruthlessly clear about what good looks like. Not "what features should we ship" but "what does success look like when this feature reaches the customer, and what trade-offs are we willing to accept to get there?"

This is harder, not easier. It moves the PO closer to a strategy role and away from a syntax role. The PO who can articulate the trade-offs gets agent output that solves the right problem. The PO who cannot gets a faster version of the wrong problem solved.

What this means for the Scrum Master role

The Scrum Master role rewrites itself second. The old SM job was to remove blockers, run ceremonies, and protect the team. The new SM job is to design the human-agent collaboration agreement.

Who decides what. When agents escalate. What the Definition of Done looks like when an agent contributed. Which decisions require human override. How the team's working memory is maintained across human and agent contributors.

These are real design problems. The SM who treats them seriously becomes a high-leverage role. The SM who keeps running meetings while agents take over the delivery becomes the slowest part of the system.

What this means for engineering managers

Engineering managers who still measure team performance with velocity charts are going to lose talent. The good engineers know velocity is dead. They know the comparison is broken. They are looking for a leader who can explain a better measure.

The EM who can articulate Concept to Cash, set targets against it, and protect the team's clarity-of-intent capacity will retain the best talent. The EM who cannot will watch their best engineers leave for places where the metric matches the era.

The objections, named honestly

There are three objections to this argument that come up every time I make it. Each is partly right.

Objection 1: Story points are still useful for sprint planning even if velocity is dead. Partly true. Inside a sprint, relative estimation still helps with slicing decisions. What is dead is the comparison, especially across teams. Use story points internally if it helps. Stop reporting velocity to leadership.

Objection 2: Not every team has agents yet. True. The transition is uneven. The point is that your competitor's team probably does, or will within 12 months. If your operating model still measures the old constraint, you will lose ground without understanding why.

Objection 3: Concept to Cash is hard to measure cleanly. True. So was velocity in 2003. We learned. We will learn this one too. Begin imperfectly. The cost of measuring an approximately-right metric beats the cost of measuring an exactly-wrong one.

How to start the shift in your team this sprint

Do not announce a metric change. Announce a question change. In the next retro, ask one new question: "What slowed the time from this feature being decided to it being live in front of customers?"

Listen to the answer. Most teams will surface a non-velocity bottleneck within 15 minutes: a stakeholder decision that took two weeks, a code review queue, an approvals chain, a customer test that was scheduled three weeks late. The slowness was never in the engineering capacity. It was in the cycle.

That question opens the door to the Concept to Cash conversation without forcing the political fight over killing velocity outright. The metric retires itself once the team starts naming the real bottleneck.

Last reviewed: 6 June 2026 by Prashant Shinde, Founder, ICAgile accredited and HRD Corp Train-the-Trainer certified. Read more: AI-native Agile primer · Glossary.