---
title: "#evals articles | Hashnode"
description: "Explore 39 articles tagged #evals on Hashnode. Tutorials, guides, and how-tos from the community."
url: "https://hashnode.com/tag/evals"
---

Tag

# #evals

39 posts · 0 followers

[Abdulwaheed A A](https://hashnode.com/@habiz) · [2d ago](https://hashnode.com/posts/why-evals-are-the-new-a-b-tests/6afc204b9fb1f09fe0725ae3)

### [Why evals are the new A/B tests](https://leveragepoints.io/why-evals-are-the-new-a-b-tests?utm_source=hashnode&utm_medium=feed)

For a long time, A/B test was one of the methods by which product teams answered questions around what good looks like. You shipped two versions, waited for traffic, and read the result. It worked wel

5 min read · [leveragepoints.io](https://leveragepoints.io/why-evals-are-the-new-a-b-tests?utm_source=hashnode&utm_medium=feed)

[ZyVOP](https://hashnode.com/@zyvop) · [5d ago](https://hashnode.com/posts/why-llm-apps-fail-in-production-six-failure-modes-and-how-to-prevent-them-felel/6af7cd349ba2b79e953d06cb)

### [Why LLM Apps Fail in Production: Six Failure Modes and How to Prevent Them](https://blog.zyvop.com/why-llm-apps-fail-in-production-six-failure-modes-and-how-to-prevent-them-felel?utm_source=hashnode&utm_medium=feed)

Rate limits, broken JSON, runaway costs, silent regressions, blind spots and prompt injection: what goes wrong after the demo works, and the fixes that hold up.

13 min read · [blog.zyvop.com](https://blog.zyvop.com/why-llm-apps-fail-in-production-six-failure-modes-and-how-to-prevent-them-felel?utm_source=hashnode&utm_medium=feed)

[2](https://hashnode.com/posts/why-llm-apps-fail-in-production-six-failure-modes-and-how-to-prevent-them-felel/6af7cd349ba2b79e953d06cb)

[AgentOps Lab](https://hashnode.com/@moneyprinterops) · [6d ago](https://hashnode.com/posts/eval-case-brief-before-you-score-the-agent-run/6af67af723037bb26d25a175)

### [Write an Eval Case Brief Before You Score the Agent Run](https://moneyprinterops.hashnode.dev/eval-case-brief-before-you-score-the-agent-run?utm_source=hashnode&utm_medium=feed)

Don't dump a failure into a blank chat and ask for a score. Write an eval case brief first: one behavior, sanitized inputs, and pass/fail signals.

7 min read · [moneyprinterops.hashnode.dev](https://moneyprinterops.hashnode.dev/eval-case-brief-before-you-score-the-agent-run?utm_source=hashnode&utm_medium=feed)

[Gary Stafford](https://hashnode.com/@garystafford) · [Nov 9](https://hashnode.com/posts/recursive-self-improvement-meets-a-strong-baseline-agent/6af1f1bb522f9a5a6377fdaf)

### [Recursive Self-Improvement Meets a Strong Baseline Agent](https://blog.latentthoughts.com/recursive-self-improvement-meets-a-strong-baseline-agent?utm_source=hashnode&utm_medium=feed)

Experiments with OpenAI agents examine correctness headroom, efficiency gains, and whether autonomous revision earns its added complexity.

16 min read · [blog.latentthoughts.com](https://blog.latentthoughts.com/recursive-self-improvement-meets-a-strong-baseline-agent?utm_source=hashnode&utm_medium=feed)

[Adarsh Divakaran](https://hashnode.com/@adarshd) · [Nov 9](https://hashnode.com/posts/serpapi-achieves-58-56-accuracy-on-seal-0-and-66-93-on-seal-hard/6af1ba8d41ed17fc85acc498)

### [SerpApi Achieves 58.56% Accuracy on SEAL-0 and 66.93% on SEAL-Hard](https://serpapi.hashnode.dev/serpapi-achieves-58-56-accuracy-on-seal-0-and-66-93-on-seal-hard?utm_source=hashnode&utm_medium=feed)

SerpApi achieved 58.56% accuracy on SEAL-0 and 66.93% on SEAL-Hard, pairing web search with page extraction to help an AI agent tackle tough factual questions.

4 min read · [serpapi.hashnode.dev](https://serpapi.hashnode.dev/serpapi-achieves-58-56-accuracy-on-seal-0-and-66-93-on-seal-hard?utm_source=hashnode&utm_medium=feed)

[Dan Holloran](https://hashnode.com/@grimicorn) · [Nov 1](https://hashnode.com/posts/llm-evals-start-as-thirty-test-cases-not-a-platform/6ae71f819ac86f1f0e22a0ac)

### [LLM Evals Start as Thirty Test Cases, Not a Platform](https://grimicorn.hashnode.dev/llm-evals-start-as-thirty-test-cases-not-a-platform?utm_source=hashnode&utm_medium=feed)

You tweak a system prompt to fix one bad answer a user reported. You try it on that example, it looks great, you ship. Two days later someone notices the feature now ignores the output format half the

5 min read · [grimicorn.hashnode.dev](https://grimicorn.hashnode.dev/llm-evals-start-as-thirty-test-cases-not-a-platform?utm_source=hashnode&utm_medium=feed)

[1](https://hashnode.com/posts/llm-evals-start-as-thirty-test-cases-not-a-platform/6ae71f819ac86f1f0e22a0ac)

[Gil Neto](https://hashnode.com/@gilneto8) · [Oct 29](https://hashnode.com/posts/my-second-brain-scored-88-and-still-told-me-something-false/6ae312a5236f3333d940fc28)

### [My second brain scored 88% and still told me something false](https://blog.gil-neto.com/my-second-brain-scored-88-and-still-told-me-something-false?utm_source=hashnode&utm_medium=feed)

On 21 October a group of researchers published a benchmark for how personal agents remember what their users want (arXiv 2610.30485). They tested several memory systems against simulated users and mea

5 min read · [blog.gil-neto.com](https://blog.gil-neto.com/my-second-brain-scored-88-and-still-told-me-something-false?utm_source=hashnode&utm_medium=feed)

[Adarsh Divakaran](https://hashnode.com/@adarshd) · [Oct 19](https://hashnode.com/posts/serpapi-achieves-97-6-accuracy-on-simpleqa-verified/6ad5e78eed098d762d464e01)

### [SerpApi Achieves 97.6% Accuracy on SimpleQA Verified](https://serpapi.hashnode.dev/serpapi-achieves-97-6-accuracy-on-simpleqa-verified?utm_source=hashnode&utm_medium=feed)

SerpApi scored 97.6% on SimpleQA Verified. We tested Google search data and the Extract API to see how well they help AI agents answer factual questions.

4 min read · [serpapi.hashnode.dev](https://serpapi.hashnode.dev/serpapi-achieves-97-6-accuracy-on-simpleqa-verified?utm_source=hashnode&utm_medium=feed)

[Ayesha Khan](https://hashnode.com/@ayeshakhalil) · [Oct 9](https://hashnode.com/posts/believable-is-not-correct/6ac94bd0659fb545e9859785)

### [Believable Is Not Correct](https://ayeshakoder.hashnode.dev/believable-is-not-correct?utm_source=hashnode&utm_medium=feed)

Why I asked what problem an AI agent should solve first, and how metrics, monitoring and guardrails turn believable output into correct output.

7 min read · [ayeshakoder.hashnode.dev](https://ayeshakoder.hashnode.dev/believable-is-not-correct?utm_source=hashnode&utm_medium=feed)

[Oliver Zehentleitner](https://hashnode.com/@oliver-zehentleitner) · [Oct 5](https://hashnode.com/posts/the-model-changed-my-skill-didn-t-the-score-still-dropped/6ac3cdcb13d6c7ff530ab3f4)

### [The Model Changed. My Skill Didn't. The Score Still Dropped.](https://blog.technopathy.club/the-model-changed-my-skill-didn-t-the-score-still-dropped?utm_source=hashnode&utm_medium=feed)

What agent evals taught me about moving model floors, noisy LLM judges, and treating the evaluator as part of the instrument

9 min read · [blog.technopathy.club](https://blog.technopathy.club/the-model-changed-my-skill-didn-t-the-score-still-dropped?utm_source=hashnode&utm_medium=feed)
