Agile and Scrum have been around for over two decades. Most companies still have not made them work end to end. Now AI is raising the bar again, and the gap between teams that deliver value and teams that perform process is about to get very visible.
The problem: Agile that never actually shipped
Many organisations are still at the beginning of the journey. They never learned to run Agile and Scrum end to end, vertically and horizontally. Instead, they kept the old way of working and put a new label on it.
Common symptoms:
- A one year backlog defined up front, with no goals or business value attached
- Micromanagement that simply renamed itself
- The same dependencies, handoffs, integration pain, features-factory, and approval layers as before
- Value delivered once a quarter or once a year
- A lot of layers to move information between teams
This is "fake Agile": the same processes, dependencies, and costs, with a new flag on top.
Why AI makes this harder, not easier
If a company is still struggling with a framework from 20 years ago, the arrival of "Agile on steroids" powered by AI agents is not a relief. It is pressure from four directions - clients, boards, employees and end-users.
Clients. They will expect fast, high-quality value on a modern development stack that is easy to maintain and migrate. Teams locked into old, hard-to-change stacks will lose traction on the market this year or next.
Boards and owners. Upskilling, cost, and a changing business model all land on their desk at once. The temptation is to add new layers and positions that mimic the classic org chart, instead of going lean. Meanwhile, the cost and speed of building products is dropping 10x to 100x.
Employees. In fake Agile teams, people stop growing. Without time to practise, even an AI certificate quickly becomes useless. If you cannot build commercial products with what you learn, you will lose your edge or move somewhere you can thrive.
End users. In an AI-enabled environment, you can deliver value daily, resolve issues daily, and build a relationship with users daily.
The risk of doing nothing
If your company could not make Agile work, what are the odds it succeeds with AI: 20%, 10%, or 0%? The likely path is a slow decline, with layoffs every year and a shrinking market position.
The "classic management layer" tends to repeat the old pattern. They will call themselves an "AI-first company" the same way they once called themselves an "agile company." Different face, same processes, dependencies, micromanagement, low value, and high cost underneath.
What to do instead
- Run a hard reality check. Name where you actually are, not where the slide deck says you are.
- Go lean. Flatten the org around DRIs and ICs, supported by AI, instead of adding layers and new positions.
- Onboard the right people. Hire motivated people who are familiar with AI and ready to transform the way you deliver.
- Create space to build. Give teams room to ship real, commercial products, not just collect certificates.
- Measure value, not motion. Track value delivered to end users, ideally on a weekly or daily cadence.
Takeaways
- Most "Agile transformations" never finished, and the symptoms are easy to spot.
- AI raises expectations for clients, boards, employees, and users at the same time.
- A new label will not close the gap. Honest assessment, lean structure, and motivated builders will.
If your transformation still feels like a slogan, start with the reality check. Everything else follows from it.