Developer shares three verification techniques for getting AI agents to finish work
11/20/2026 — 11/20, 05:17·1 sources (all)·1 reports (all)
Key takeaways
- A practitioner uses an agent that drew a line labelled 4.5 metres while its endpoint landed at 3.5 metres to argue that you must define how…
Background & analysis
On November 20, 2026, the DEV Community AI section published a practitioner write-up titled “Three Verification Tips to Get AI Agents to Actually Finish the Work.” The author opens with a real task he handed to an AI agent: adding a 4.5-meter line to a drawing. The agent labeled the line “4.5 m,” yet the endpoint landed exactly at the end of the old 3.5-meter line. The label was accurate, but the drawing contradicted it.
The author points out that this kind of mismatch is easy to catch on a drawing, while the same class of problem becomes far harder to spot in a research report. As a second example, a sentence in a report claims that a new process cut costs by 30%. Clicking the citation shows that the source says costs fell by up to 30% in a single pilot. The link works, but the sentence does not match the source it points to. Both cases point to the same lesson: before delegating work to an AI agent, the person assigning the task needs to know in advance how to judge whether the result is correct.
To close that gap, the author reuses three skills. One of them, Delivery Roadmap, breaks a task into smaller steps that come with checkpoints, and it calls for inserting verifiable intermediate checks in steps such as drawing and citation verification, so the agent is verified along the way rather than only at the end. The piece remains a single first-hand account, and no further coverage or follow-up has appeared so far.
AI-generated from 1 reports · updated 2 days ago
- Heat index
- 25
- All sources
- 1
- All reports
- 1
- First seen
- 2 days ago
Reports on this story headlines open the original
The author recounts an agent that labeled a line '4.5 m' while its endpoint landed on the old 3.5 m mark, showing how a result can look right and still be wrong. He now asks how a result will be verified before handing over a task, and reuses three skills, including a Delivery Roadmap that splits work into small steps with a check for each one.
DEV Community · AIAI score 52
Other stories people are talking about
- Heat index 76Developer maps five prompt failure modes for on-device Android agents1 sources
- Heat index 47Guide: spotting when AI coding agents weaken or delete your tests1 sources
- Heat index 1143RisingAnthropic Reaches $1.6 Trillion Market Value in First Trading Week After IPO21 sources
- Heat index 1011Anthropic Mathematician Says Claude Formalized the Generalized Riemann Hypothesis21 sources
- Heat index 622Anthropic IPO total raise reaches $93.5 billion after greenshoe exercise20 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion

