{"rewrite":{"id":"r_0e390b914c61d64ff7765037","clusterId":"c_b70ff44e420e5a77d2398324","slug":"langfuse-lets-you-watch-ai-agents-work-and-track-token-use-retroactively","model":"deepseek-v4-flash","headline":"Langfuse Lets You Watch AI Agents Work and Track Token Use Retroactively","summary":"Langfuse is an open-source, self-hostable AI engineering platform that records the full execution path of LLM calls and AI agents as traces. From a browser, users can review which LLM was called, which tools executed, what was searched, processing time, token usage, and estimated costs. It integrates with OpenTelemetry, LangChain, the Google Gen AI SDK, and LiteLLM, and was part of Y Combinator's W23 batch.","whyItMatters":"The tool makes AI agent internals visible after the fact, turning a black-box token spend into a browsable record of every search, tool call, and model query.","webCardHtml":"\u003cp\u003eThe setup walkthrough starts with creating an organization and project, then issuing API keys. The writer notes a bug where the screen does not transition properly after project creation, advising users to click Projects in the side menu manually. The secret key appears only once, so it must be copied and saved at creation.\u003c/p\u003e\u003cp\u003eTo demonstrate, the writer integrated Langfuse with Google Gemini through the Google Gen AI SDK and Observability for Google Gemini Models with Langfuse Integration. A deliberately redundant prompt asked the agent to re-check weather information with a different search keyword. The agent compared multiple sources and concluded a foldable umbrella was needed for Tokyo, despite clouds and sunny periods, because sudden rain was possible from mid-afternoon to evening.\u003c/p\u003e","blueskyPost":"Langfuse records the full execution path of an AI agent: which LLM was called, which tools ran, what was searched, and how many tokens were used. Open-source and self-hostable, viewable from a browser.","twitterPost":"Langfuse is an open-source, self-hostable platform that traces AI agent execution: LLM calls, tool use, searches, processing time, token usage, and estimated costs, all reviewable later in a browser.","threadsPost":null,"newsletterBlurb":"Langfuse is an open-source, self-hostable AI engineering platform that records LLM and AI agent execution as traces. It lets developers check which models were called, which tools ran, what was searched, processing time, token usage, and estimated costs from a browser. The review demonstrates integration with Google Gemini and notes a screen transition bug during project creation.","attributionJson":"[{\"source\":\"GIGAZINE\",\"url\":\"https://gigazine.net/news/20260823-langfuse/\",\"title\":\"Langfuse: An Open-Source, Self-Hostable Tool That Lets You See How AI Agents Actually Work and Use Tokens Retroactively\"}]","lintFlagsJson":null,"lintHits":0,"costUsd":0,"inputTokens":4644,"outputTokens":599,"status":"published","repairAttempts":0,"nextRepairAt":null,"factsAttemptedAt":1787994596,"createdAt":"2026-08-29T09:01:24.000Z","publishedAt":"2026-08-29T09:01:45.000Z","updatedAt":"2026-08-29T09:01:24.000Z"},"cluster":{"id":"c_b70ff44e420e5a77d2398324","canonicalTitle":"AIエージェントが実際にどのように動いてトークンを使っているかが後から見てわかるようにできる「Langfuse」、オープンソースでセルフホスト可能","representativeArticleId":"a_324028d6d23dcf19e7c54ef5","sourceCount":1,"writtenSourceCount":1,"writeAttempts":0,"isSolo":true,"entitiesJson":"{\"anime_titles\":[],\"manga_titles\":[],\"work_titles\":[],\"studios\":[],\"people\":[],\"type\":\"news\",\"domain\":\"other\",\"is_roundup\":false}","contentType":"news","status":"published","firstSeenAt":"2026-08-23T14:00:00.000Z","lastSeenAt":"2026-08-23T14:00:00.000Z","updatedAt":"2026-08-29T09:01:45.000Z"},"attribution":[{"source":"GIGAZINE","url":"https://gigazine.net/news/20260823-langfuse/","title":"AIエージェントが実際にどのように動いてトークンを使っているかが後から見てわかるようにできる「Langfuse」、オープンソースでセルフホスト可能"}],"entities":{"anime_titles":[],"manga_titles":[],"work_titles":[],"studios":[],"people":[],"type":"news","domain":"other","is_roundup":false},"keyFacts":null}