--- title: How do I monitor and trace AI agents running in Production description: Running the same AI agent task 720 times and planting one bad tool result exposed a gap traditional logging can't catch. In 5 of 5 sessions with a corrupted... canonical: https://daily.dev/posts/how-do-i-monitor-and-trace-ai-agents-running-in-production-wulrtjnmd twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: How do I monitor and trace AI agents running in Production | daily.dev og:description: Running the same AI agent task 720 times and planting one bad tool result exposed a gap traditional logging can't catch. In 5 of 5 sessions with a corrupted... og:url: https://daily.dev/posts/how-do-i-monitor-and-trace-ai-agents-running-in-production-wulrtjnmd og:image: https://api.daily.dev/og/posts/WuLRtjnMd.png og:image:alt: How do I monitor and trace AI agents running in Production 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. # How do I monitor and trace AI agents running in Production **[DigitalOcean Community](https://daily.dev/sources/do_community)** · 20 min read · 0 upvotes · 0 comments ## Summary Running the same AI agent task 720 times and planting one bad tool result exposed a gap traditional logging can't catch. In 5 of 5 sessions with a corrupted tool result and no user pushback, every tool call succeeded and the agent still gave a wrong answer. A strict log with just the request and final answer never found the bad call in 15 tries. Full tool-call capture (arguments, results) let a debugger find the fault in 9 to 15 of 15 tries, rising to 15 of 15 with a reference lookup added. The 15 counted sessions cost about $1.59 in platform-reported run cost on DigitalOcean Managed Agents. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Why didn't my AI agent's logs show any errors even though it gave a wrong answer to the user? Traditional logs only record errors, but a wrong answer from bad tool data finishes cleanly with no exception. In a controlled test, 5 of 5 sessions fed one corrupted tool result had every tool call succeed while the final answer was still wrong. A plain log rarely shows which record was bad unless the agent's text happens to repeat the data verbatim. _Anyone debugging silent AI agent failures can track approaches like full-result logging on daily.dev._ ### What should I log for each tool call to debug AI agent failures effectively? Log the tool name, arguments, the full result (not just success or failure), the model ID, and the prompt version for every call. For large outputs, store a hash plus a pointer to the full result instead. Testing showed full tool results alone let an automated debugger find a planted bad call in 9 to 11 of 15 tries, rising to 15 of 15 when a reference lookup was added. _Developers designing agent observability pipelines can compare logging strategies on daily.dev._ ### Does re-running an AI agent at temperature 0 reproduce the same output every time? Only partly, and it depends on the server. On a dedicated vLLM server, temperature 0 gave the identical answer across all 400 runs. On a hosted endpoint at temperature 0, the tool-call path and final answer repeated, but all 40 generated response texts still differed in wording. Re-running also can't recover a one-time bad tool result, since a second lookup returns clean data. _Teams relying on deterministic LLM outputs can follow findings like this on daily.dev._ ## Similar posts on daily.dev - [How we eliminated $1 million a year of wasted AI agent spend in one hour](https://daily.dev/posts/how-we-eliminated-1-million-a-year-of-wasted-ai-agent-spend-in-one-hour-sl4y8p7za) · databricks · 1 upvotes · 0 comments - [Session Traces and Cost Controls Help Diagnose AI Agent Failures](https://daily.dev/posts/session-traces-and-cost-controls-help-diagnose-ai-agent-failures-cphogfesa) · InfoQ · 0 upvotes · 0 comments - [AI Agent Observability: Logging, Tracing, and Debugging Explained](https://daily.dev/posts/ai-agent-observability-logging-tracing-and-debugging-explained-95izimukh) · Machine Learning Mastery · 1 upvotes · 0 comments - [How to Debug AI Coding Agents When They Change the Wrong Thing](https://daily.dev/posts/how-to-debug-ai-coding-agents-when-they-change-the-wrong-thing-f9znhzoc8) · Towards Data Science · 1 upvotes · 0 comments - [Optimize AI costs with Agent Observability](https://daily.dev/posts/optimize-ai-costs-with-agent-observability-lnsouoalu) · Datadog · 0 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#llm](https://daily.dev/tags/llm), [#observability](https://daily.dev/tags/observability), [#digitalocean](https://daily.dev/tags/digitalocean) [View this post on daily.dev](https://daily.dev/posts/how-do-i-monitor-and-trace-ai-agents-running-in-production-wulrtjnmd) ```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":"How do I monitor and trace AI agents running in Production","url":"https://daily.dev/posts/how-do-i-monitor-and-trace-ai-agents-running-in-production-wulrtjnmd","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/how-do-i-monitor-and-trace-ai-agents-running-in-production-wulrtjnmd"},"datePublished":"2026-10-30T21:43:36.246Z","dateModified":"2026-10-30T21:44:28.042Z","description":"Running the same AI agent task 720 times and planting one bad tool result exposed a gap traditional logging can't catch. In 5 of 5 sessions with a corrupted...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/130bd12ee9eea40b0a9557c54622a7e8?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/130bd12ee9eea40b0a9557c54622a7e8?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"DigitalOcean Community","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":"DigitalOcean Community","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/c1b9d07730e34ea388c39a498a753d6c","url":"https://daily.dev/sources/do_community"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/how-do-i-monitor-and-trace-ai-agents-running-in-production-wulrtjnmd","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,llm,observability,digitalocean","timeRequired":"PT20M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"DigitalOcean Community","item":"https://daily.dev/sources/do_community"},{"@type":"ListItem","position":3,"name":"How do I monitor and trace AI agents running in Production"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/how-do-i-monitor-and-trace-ai-agents-running-in-production-wulrtjnmd#faq","mainEntity":[{"@type":"Question","name":"Why didn't my AI agent's logs show any errors even though it gave a wrong answer to the user?","acceptedAnswer":{"@type":"Answer","text":"Traditional logs only record errors, but a wrong answer from bad tool data finishes cleanly with no exception. In a controlled test, 5 of 5 sessions fed one corrupted tool result had every tool call succeed while the final answer was still wrong. A plain log rarely shows which record was bad unless the agent's text happens to repeat the data verbatim. Anyone debugging silent AI agent failures can track approaches like full-result logging on daily.dev."}},{"@type":"Question","name":"What should I log for each tool call to debug AI agent failures effectively?","acceptedAnswer":{"@type":"Answer","text":"Log the tool name, arguments, the full result (not just success or failure), the model ID, and the prompt version for every call. For large outputs, store a hash plus a pointer to the full result instead. Testing showed full tool results alone let an automated debugger find a planted bad call in 9 to 11 of 15 tries, rising to 15 of 15 when a reference lookup was added. Developers designing agent observability pipelines can compare logging strategies on daily.dev."}},{"@type":"Question","name":"Does re-running an AI agent at temperature 0 reproduce the same output every time?","acceptedAnswer":{"@type":"Answer","text":"Only partly, and it depends on the server. On a dedicated vLLM server, temperature 0 gave the identical answer across all 400 runs. On a hosted endpoint at temperature 0, the tool-call path and final answer repeated, but all 40 generated response texts still differed in wording. Re-running also can't recover a one-time bad tool result, since a second lookup returns clean data. Teams relying on deterministic LLM outputs can follow findings like this on daily.dev."}}]} ```