Agents demo beautifully and fail quietly. Gartner expects over 40% of agentic AI projects to be cancelled by the end of 2027 — citing escalating costs, unclear value and weak risk controls. Here's what actually goes wrong.
The failure modes
- Compounding error. An agent chains steps; a 90%-reliable step run five times is only ~59% reliable end to end.
- Unbounded cost & loops. Without limits, an agent can spiral — retrying, re-planning, burning tokens.
- No guardrails. Letting output act (run code, send mail, move money) turns a hallucination into an incident.
- No evaluation. You can't improve what you can't measure, and agent trajectories are hard to score.
An agent doesn't make an unreliable workflow reliable. It makes it autonomous — which is worse.
What the survivors do
- 1.Start with one narrow, valuable task, not a general agent.
- 2.Constrain tools and permissions to the minimum the task needs.
- 3.Keep a human approving anything irreversible.
- 4.Instrument every step — inputs, outputs, cost, latency, success.
Sources
Written by ivector
Start a project →