# How do you pick an AI agent project that will not get cancelled?

Pick a job with a known cost, clear rules and 1 owner. Gartner expects over 40% of agentic AI projects to be cancelled by the end of 2027. It names 3 causes: rising costs, unclear business value and weak risk controls. Each can be checked before the build.

By Surinder Thakur, AI Product Designer and Builder. Published 17 Nov 2026. Topic: [Building an AI tool](https://www.surinder.design/work?type=building-an-ai-tool).

## How do you catch rising costs early?

Price the run cost, not only the build. Model usage grows with every task the agent handles. A pilot that is cheap at 50 tasks can hurt at 50,000.

Estimate the monthly volume first, then the cost per task. If running the agent costs more than doing the work today, stop there.

## How do you prove the business value?

Pick a task that costs money you can name today. Hours per week, errors that end in refunds, orders lost to slow replies.

A project sold as innovation has no number to defend. When budgets get cut, it goes first.

## What counts as a risk control?

3 things: rules the agent cannot cross, an undo, and a log of every step. A person decides anything outside the rules.

Without them, the first visible mistake ends the project, however good the agent is.

## Is the job worth an agent at all?

Sometimes it is not. Gartner advises pursuing agentic AI only where it delivers clear value or ROI. Some jobs need a simple automation or a better form.

Saying no early is the cheapest outcome. The 2 week Blueprint is built to reach that answer before the build spend.

## In 1 picture

20 agent ideas, narrowed by 3 checks: cost, value and risk.

20 ideas, narrowed to 8, 3, then 1 project.

Most agent ideas fail a cost, value or risk check. The one that passes all 3 gets built.

An illustration of the 3 checks, not data. The 40% figure is Gartner's.

## Sources

- [Gartner · Over 40% of agentic AI projects will be canceled by end of 2027, 2025](https://www.gartner.com/en/newsroom/press-releases/2025-06-25-gartner-predicts-over-40-percent-of-agentic-ai-projects-will-be-canceled-by-end-of-2027)

## Next questions

- [Which task in my business should get an AI agent first?](https://www.surinder.design/work/which-task-should-get-an-ai-agent-first): Pick a task that repeats every week, is easy to check, and has 1 owner.
- [Why do AI pilots stall before real use?](https://www.surinder.design/work/why-ai-pilots-stall): AI pilots stall because a demo only proves the tool can work.
- [How do you measure the ROI of an AI agent?](https://www.surinder.design/work/how-to-measure-the-roi-of-an-ai-agent): Count 4 numbers. The hours the task takes today, priced at what those hours cost.
- [What is agent washing, and how do you spot it?](https://www.surinder.design/work/what-is-agent-washing): Agent washing is selling an old chatbot or automation as an AI agent.

## Questions that lead here

- [What if the build does not pay off?](https://www.surinder.design/work/what-if-the-build-does-not-pay-off): Then you find out in the Blueprint, before the build spend.

Page: https://www.surinder.design/work/how-to-pick-an-ai-agent-project
