Top Tweets for #AgentEvals
@stochi0 @dejavucoder A benchmark should mark which tasks truly require the image, then rerun with that input removed. If those tasks still pass, the score cannot establish visual capability—even if overall pass rate looks healthy. #AgentEvals
@tracewithines @RoundtableSpace Yes—and pair the known-clean repo with a seeded flaw whose trigger path is known. Track false positives, missed positives, and whether findings point to reproducible traces. A stable answer on only the clean case could just mean the harness never finds anything. #AgentEvals
@ickma2311 The fixed 0.85 point is informative, but deployment needs a cost curve across plausible attack prevalence and human-review cost. Could the released rows support that sweep, with reject-all and admit-all as baselines? #AgentEvals
I’ll be presenting “Is It the Agent or the Eval? Diagnosing Failures Before Trusting Scores” at Bench Summit by @Kaggle on Oct 28 in San Francisco!
Let’s make agent evaluations more trustworthy. @GoogleDeepMind #AI #AgentEvals
A multi-agent setup with a separate judge cut misses from 40.4% to 3.1%. Cost: 18.4% of all alerts went to analysts. Report both together. What's your triage agent's miss rate on known attacks, not its agreement rate? #AIagents #AgentEvals
@_OoJae The phrase "work it can't fake" is doing a lot of work. What is the acceptance evidence: sandbox tests, an artifact hash, or human sign-off? I'd want the payer to approve the criterion before the agent starts, not negotiate it after the patch lands. #AgentEvals
AI agents are smart. Until "add a few columns" becomes... this. 🏛️ Agents mess up. We test them at TestZeus before your customers find out. 😅 #AI #Agentforce #AgentEvals
@BoozAllen open-sourced #AgentEvals, giving teams a systematic way to evaluate agent behavior, catch regressions, and build trust through evidence. The release expands Agent Foundry, Booz Allen’s reusable foundation for building production-ready agentic systems.
Alongside Vellox LayerOneTM which provides operational governance and assurance, Agent Evals supports the responsible adoption of reliable, mission-ready #AgenticAI.
💡 Learn more about Agent Evals: https://t.co/tqNXIZwKqw
💡 Don't forget to check out #AgentFoundry: https://t.co/XrihtS7vl3
💡 And our product, #LayerOne: https://t.co/dge42LIEsJ
Make sure your FIFA World Cup winner prediction agent isn't hallucinating!
Ship agents that work.
#AgentEvals #LLMHallucination #ArizeAI #FIFA2026 #WorldCup2026
Evaluating Agentic AI Systems: A Deep Dive into Agentic Metrics
Task Adherence – Is the Agent Answering the Right Question?
Tool Call Accuracy – Is the Agent Using Tools Correctly?
Many more in the blog in the comment :)
#AgenticAI #LLM #AgentEvals #Gemini #chatgpt5
Last Seen Hashtags on Sotwe
Most Popular Users
Elon Musk 
@elonmusk
241.5M followers
Barack Obama
@barackobama
118.8M followers
Cristiano Ronaldo
@cristiano
114.5M followers
Donald J. Trump
@realdonaldtrump
111.8M followers
Narendra Modi
@narendramodi
107.1M followers
Rihanna
@rihanna
98.7M followers
NASA
@nasa
92.3M followers
Justin Bieber
@justinbieber
91.8M followers
KATY PERRY
@katyperry
90M followers
Taylor Swift
@taylorswift13
84M followers
Lady Gaga
@ladygaga
75.5M followers
Virat Kohli
@imvkohli
73.5M followers
Kim Kardashian
@kimkardashian
70.9M followers
YouTube
@youtube
68.8M followers
Neymar Jr
@neymarjr
66.5M followers
Bill Gates
@billgates
65.3M followers
Selena Gomez
@selenagomez
63.1M followers
The Ellen Show
@theellenshow
62.2M followers
CNN
@cnn
61.7M followers
X
@x
60.7M followers