Lemma has raised $2.3 million in pre-seed funding to find and fix silent failures in AI agents.<\/p>\n
Investors from OpenAI, xAI, Meta, and DoorDash joined the round, along with Matrix and Y Combinator.<\/p>\n
Lemma reviews more than one million agent traces daily to identify the main reasons for failures.<\/p>\n
Jerry Zhang<\/a> and Cole Gawin<\/a> met as freshmen in USC\u2019s startup incubator before building agents at separate companies. Both faced the same issue: agents passed all tests but failed in real-world use, with no log data indicating the problem.<\/p>\n In these situations, there are no crashes or error messages. Customers just stop coming back, so development teams have no way to see what went wrong. Lemma<\/a> wants to solve this problem with its $2.3 million pre-seed funding.\u00a0<\/p>\n The round included Matrix, Y Combinator, Liquid 2 Ventures, Vermilion Cliffs Ventures, Irregular Expressions, Cervin Ventures, Comma Capital, Position Ventures, and Eight Capital, along with angel investors from OpenAI<\/a>, xAI, Meta, and DoorDash<\/a>.\u00a0<\/p>\n \u201cCole and I started Lemma because we experienced the pain of building AI agents firsthand. We kept running into the same problem: agents would appear to work, but the results weren\u2019t reliable enough in production,\u201d said Zhang, co-founder of Lemma.\u00a0<\/p>\n How Lemma works\u00a0<\/p>\n Founded in 2025 and based in San Francisco, Lemma monitors live agent traffic to detect issues such as infinite loops, failed tool calls, and misinterpretations of intent.\u00a0<\/p>\n The platform reviews over one million agent traces each day, identifies root causes, and suggests fixes directly in the codebase, making detection and resolution faster. Its launch video shows this by recreating the viral restaurant scene from Obsession, highlighting how agents can look like they work until something goes wrong.<\/p>\n The market is crowded and changing quickly. Dash0 became a unicorn in March 2026 after raising $110 million from<\/a> Balderton, with clients like Zalando and Taco Bell. FriskAI raised $3.6 million in pre-seed funding from MaC Venture Capital<\/a> to monitor agent behavior. Tsuga, a Paris-based company founded by former Datadog engineers, raised $35 million in Series A funding<\/a> to keep telemetry inside customers\u2019 cloud environments.\u00a0<\/p>\n Most competitors offer dashboards and evaluation scores, but Lemma stands out by offering an all-in-one solution that finds and fixes issues, helping teams that do not have their own evaluation resources.<\/p>\n Investors and next steps<\/p>\n The funding will help Lemma develop its product and expand its detection tools, focusing on seed-to-Series B teams running agents at real scale. The AI agent observability market is worth about $0.9 billion in 2026<\/a> and could grow to $14 billion by 2035, at an annual growth rate of 35.6%.<\/p>\n \u201cWe\u2019re in the very early innings of improving agent quality, and I\u2019m excited to see the Lemma founders tackle it with a focus on silent failures and automated resolution,\u201d said Ilya Sukhar<\/a>, general partner at Matrix.<\/p>\n \u201cExisting tools tell you what broke after the fact. Lemma tells you whether the agent actually did its job, catching silent failures before users churn. Jerry and Cole hit the same wall building agents at two unrelated companies, so they understand this problem from the inside. They bring infectious energy and obsessive problem-solving to catching the silent failures that erode trust, and they\u2019re moving faster than anyone I\u2019ve seen,\u201d added Ashley Smith<\/a>, founder of Vermilion Cliffs Ventures.<\/p>\n Every incumbent racing to bolt an \u201cAI agent\u201d feature onto an existing dashboard is assuming customers will keep tolerating failures they can\u2019t see. Lemma\u2019s founders are counting on customers reaching their limit first. Whether a two-person team can capitalise on that before Databricks, Cisco, or Datadog build the same thing internally is the open question, and probably the one Lemma\u2019s next round will answer.<\/p>\n","protected":false},"excerpt":{"rendered":"Lemma has raised $2.3 million in pre-seed funding to find and fix silent failures in AI agents. Investors…\n","protected":false},"author":2,"featured_media":138720,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[10],"tags":[24,3946,66,1162,156,2899],"class_list":["post-138719","post","type-post","status-publish","format-standard","has-post-thumbnail","category-xai","tag-ai","tag-funding","tag-news","tag-startups","tag-us","tag-xai"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/138719","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=138719"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/138719\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/138720"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=138719"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=138719"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=138719"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}