Unverified50% confidenceFactExact time
According to one benchmark test that had agents complete tasks according to rules and databases in retail and airline customer service scenarios, the best-performing model at the time had an accuracy of less than half.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/1/2026
First Seen
Valid until: 1/30/2027
Sources
What Is an AI Agent? The Essential Difference Between Agents and Chatbots
youtubeRadar 360 AI10/31/2026
Related Entities
Related Claims
UnverifiedAbsolute scores on conventional agentic benchmarks are generally low; even leading models score less than 20% on the legal work test, because the tasks themselves simulate work that would take human experts hours or even days to complete.69% similarUnverifiedEven when the correct value was retrieved, agents acted on it only 55% of the time, exposing a gap between retrieving information and acting on it.68% similarUnverifiedIn 10%–25% of cases, reasoning about the hypothetical grader pulled the agent's work away from the user's original specification.68% similarUnverifiedFor traditional customer service automation systems based on keyword-matching rules, after one month the intent recognition accuracy typically stalls at around 65%, with a task completion rate of only 40%.68% similarUnverifiedThe user pointed out that these models achieve high scores on benchmarks yet frequently fail at the simplest tasks.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1122453API
curl https://kongchang.com/api/v1/knowledge/claims/1122453MCP
get_claim(id=1122453)