--- title: Episode #302: Constructing and Judging Modern Agentic... description: Quality Engineer Suneet Malhotra discusses building and evaluating agentic AI architectures in Python. Key topics include specification-enrichment loops that... canonical: https://daily.dev/posts/episode-302-constructing-and-judging-modern-agentic-workflows-the-real-python-podcast-brenwj0z0 twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Episode #302: Constructing and Judging Modern Agentic Workflows – The Real Python Podcast | daily.dev og:description: Quality Engineer Suneet Malhotra discusses building and evaluating agentic AI architectures in Python. Key topics include specification-enrichment loops that... og:url: https://daily.dev/posts/episode-302-constructing-and-judging-modern-agentic-workflows-the-real-python-podcast-brenwj0z0 og:image: https://api.daily.dev/og/posts/Brenwj0z0.png og:image:alt: Episode #302: Constructing and Judging Modern Agentic Workflows – The Real Python Podcast og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Episode #302: Constructing and Judging Modern Agentic Workflows – The Real Python Podcast **[Real Python](https://daily.dev/sources/rpython)** · 3 min read · 1 upvotes · 0 comments ## Summary Quality Engineer Suneet Malhotra discusses building and evaluating agentic AI architectures in Python. Key topics include specification-enrichment loops that improve LLM agent systems, using an LLM as a judge to evaluate agent outputs, and applying Cohen's kappa to measure inter-rater agreement between LLM judgments. The conversation also covers failure modes in agentic workflows, cross-layer observability for LLM-assisted test automation, and low-cost ways to experiment with these techniques. Malhotra links to several peer-reviewed papers and GitHub projects for hands-on exploration. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is specification enrichment in the context of agentic AI systems? Specification enrichment is a technique for improving LLM agent systems by iteratively refining and saving the specifications that guide agent behavior, creating a feedback loop that strengthens how well the agent performs its task. Suneet Malhotra, a quality engineering researcher with 18 years of experience, developed this approach and published a reference implementation and empirical evaluation as version 28 for EISEJ submission on Zenodo. _Engineers refining agent specifications can follow ongoing agentic AI practices on daily.dev._ ### How can Cohen's kappa be used to evaluate LLM-as-judge systems? Cohen's kappa can measure the agreement between judgments made by an LLM acting as a judge, providing a statistical way to validate whether an LLM judge's evaluations are consistent and reliable rather than random. This metric, borrowed from inter-rater reliability statistics, helps quality engineers quantify trust in automated LLM-based evaluation within agentic architectures. _Teams validating LLM-as-judge reliability can track evaluation methodology discussions on daily.dev._ ## Similar posts on daily.dev - [Building an agentic SDLC with a QA engineering mindset](https://daily.dev/posts/building-an-agentic-sdlc-with-a-qa-engineering-mindset-8i8iowcsg) · Stack Overflow Blog · 0 upvotes · 0 comments - [How to Build Agentic AI Workflows](https://daily.dev/posts/how-to-build-agentic-ai-workflows-cubaycvo2) · freeCodeCamp · 2 upvotes · 0 comments - [Episode \#291: Reassessing the LLM Landscape & Summoning Ghosts – The Real Python Podcast](https://daily.dev/posts/episode-291-reassessing-the-llm-landscape-summoning-ghosts-the-real-python-podcast-tlxonslv8) · Real Python · 1 upvotes · 0 comments --- Tags: [#python](https://daily.dev/tags/python), [#testing](https://daily.dev/tags/testing), [#ai-agents](https://daily.dev/tags/ai-agents) [View this post on daily.dev](https://daily.dev/posts/episode-302-constructing-and-judging-modern-agentic-workflows-the-real-python-podcast-brenwj0z0) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Episode #302: Constructing and Judging Modern Agentic Workflows – The Real Python Podcast","url":"https://daily.dev/posts/episode-302-constructing-and-judging-modern-agentic-workflows-the-real-python-podcast-brenwj0z0","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/episode-302-constructing-and-judging-modern-agentic-workflows-the-real-python-podcast-brenwj0z0"},"datePublished":"2026-07-10T12:13:40.329Z","dateModified":"2026-09-14T06:59:53.051Z","description":"Quality Engineer Suneet Malhotra discusses building and evaluating agentic AI architectures in Python. Key topics include specification-enrichment loops that...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/c85af51d5b96deb7153ea95c6c5308b1?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/c85af51d5b96deb7153ea95c6c5308b1?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Real Python","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Real Python","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/rpython","url":"https://daily.dev/sources/rpython"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/episode-302-constructing-and-judging-modern-agentic-workflows-the-real-python-podcast-brenwj0z0","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"python,testing,ai-agents","timeRequired":"PT3M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Real Python","item":"https://daily.dev/sources/rpython"},{"@type":"ListItem","position":3,"name":"Episode #302: Constructing and Judging Modern Agentic Workflows – The Real Python Podcast"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/episode-302-constructing-and-judging-modern-agentic-workflows-the-real-python-podcast-brenwj0z0#faq","mainEntity":[{"@type":"Question","name":"What is specification enrichment in the context of agentic AI systems?","acceptedAnswer":{"@type":"Answer","text":"Specification enrichment is a technique for improving LLM agent systems by iteratively refining and saving the specifications that guide agent behavior, creating a feedback loop that strengthens how well the agent performs its task. Suneet Malhotra, a quality engineering researcher with 18 years of experience, developed this approach and published a reference implementation and empirical evaluation as version 28 for EISEJ submission on Zenodo. Engineers refining agent specifications can follow ongoing agentic AI practices on daily.dev."}},{"@type":"Question","name":"How can Cohen's kappa be used to evaluate LLM-as-judge systems?","acceptedAnswer":{"@type":"Answer","text":"Cohen's kappa can measure the agreement between judgments made by an LLM acting as a judge, providing a statistical way to validate whether an LLM judge's evaluations are consistent and reliable rather than random. This metric, borrowed from inter-rater reliability statistics, helps quality engineers quantify trust in automated LLM-based evaluation within agentic architectures. Teams validating LLM-as-judge reliability can track evaluation methodology discussions on daily.dev."}}]} ```