--- title: How Apollo Uses Deep Agents and LangSmith for GTM AI description: Apollo rebuilt its AI Assistant on LangChain's Deep Agents framework, moving away from a LangGraph supervisor-based multi-agent architecture to a flat,... canonical: https://daily.dev/posts/how-apollo-uses-deep-agents-and-langsmith-for-gtm-ai-2nzctoe3y twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: How Apollo Uses Deep Agents and LangSmith for GTM AI | daily.dev og:description: Apollo rebuilt its AI Assistant on LangChain's Deep Agents framework, moving away from a LangGraph supervisor-based multi-agent architecture to a flat,... og:url: https://daily.dev/posts/how-apollo-uses-deep-agents-and-langsmith-for-gtm-ai-2nzctoe3y og:image: https://api.daily.dev/og/posts/2NzCToe3Y.png og:image:alt: How Apollo Uses Deep Agents and LangSmith for GTM AI 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 Apollo Uses Deep Agents and LangSmith for GTM AI **[LangChain](https://daily.dev/sources/langchain)** · 6 min read · 1 upvotes · 0 comments ## Summary Apollo rebuilt its AI Assistant on LangChain's Deep Agents framework, moving away from a LangGraph supervisor-based multi-agent architecture to a flat, skill-driven model. The new architecture dynamically selects from a library of GTM skills (prospecting, enrichment, outreach, analytics) based on user intent, reducing confirmation prompts and cutting development time by 80-85%. Apollo pairs this with LangSmith for tracing and a six-layer evaluation framework called AI Watchtower, covering quality rubrics, golden tests, live traces, aggregated metrics, weekly pulse reports, and in-product feedback. The assistant has expanded beyond the UI to a headless API and MCP server, now powering integrations with Claude, ChatGPT, and Perplexity, with 40k+ teams already building on the Apollo MCP. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Why did Apollo switch from a LangGraph supervisor architecture to Deep Agents for its AI assistant? Apollo moved from a LangGraph supervisor-based multi-agent system to LangChain's Deep Agents because the deterministic graph structure required writing and wiring a new sub-agent for every use case, causing frequent user confirmation prompts. Deep Agents let the assistant dynamically select from a skills library based on stated intent, cutting development time from initial build to launch by roughly 80-85% and reducing confirmation prompts. _Engineers weighing agent framework migrations can find real-world architecture writeups like this on daily.dev._ ### Why did Apollo choose Deep Agents over Anthropic's Claude Agent SDK for its GTM AI assistant? Apollo chose Deep Agents over the Claude Agent SDK because Claude SDK would have locked the assistant into Anthropic models only, while Deep Agents preserves model neutrality across multiple LLM vendors. Apollo's engineering team viewed vendor lock-in as closing doors permanently, making model flexibility the deciding factor. _Teams comparing agent SDKs for vendor lock-in tradeoffs follow these evaluations on daily.dev._ ### What is Apollo's AI Watchtower evaluation framework for AI agents? AI Watchtower is Apollo's six-layer evaluation strategy built on LangSmith, covering pre-launch quality rubrics scored by two reviewers, end-to-end golden-scenario tests run on every PR, live trace debugging, aggregated live insights with drill-downs on quality drops, a weekly pulse report by cohort, and customer thumbs-up/down feedback that triggers a P1 incident if thumbs-down exceeds 8% over a rolling 7-day window. _Anyone designing eval pipelines for production AI agents can track similar approaches on daily.dev._ ## Similar posts on daily.dev - [Debugging Deep Agents with LangSmith](https://daily.dev/posts/debugging-deep-agents-with-langsmith-v1oyf16pl) · LangChain · 1 upvotes · 0 comments - [How I Built an Agentic AI Research Assistant Using LangGraph and Large Language Models](https://daily.dev/posts/how-i-built-an-agentic-ai-research-assistant-using-langgraph-and-large-language-models-dg5wempfs) · Medium · 0 upvotes · 0 comments - [How to Build a Multi-Agent AI System with LangGraph, MCP, and A2A \[Full Book\]](https://daily.dev/posts/how-to-build-a-multi-agent-ai-system-with-langgraph-mcp-and-a2a-full-book--qyzb80uxm) · freeCodeCamp · 3 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#mcp](https://daily.dev/tags/mcp), [#langgraph](https://daily.dev/tags/langgraph), [#langsmith](https://daily.dev/tags/langsmith), [#llm-observability](https://daily.dev/tags/llm-observability) [View this post on daily.dev](https://daily.dev/posts/how-apollo-uses-deep-agents-and-langsmith-for-gtm-ai-2nzctoe3y) ```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 Apollo Uses Deep Agents and LangSmith for GTM AI","url":"https://daily.dev/posts/how-apollo-uses-deep-agents-and-langsmith-for-gtm-ai-2nzctoe3y","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/how-apollo-uses-deep-agents-and-langsmith-for-gtm-ai-2nzctoe3y"},"datePublished":"2026-07-21T18:31:02.711Z","dateModified":"2026-09-14T09:10:16.075Z","description":"Apollo rebuilt its AI Assistant on LangChain's Deep Agents framework, moving away from a LangGraph supervisor-based multi-agent architecture to a flat,...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/aa9360d8ad602da58e935ddb58fafa83?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/aa9360d8ad602da58e935ddb58fafa83?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"LangChain","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":"LangChain","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/0f4f2629e0724a849823f8cd0d913e13","url":"https://daily.dev/sources/langchain"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/how-apollo-uses-deep-agents-and-langsmith-for-gtm-ai-2nzctoe3y","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,mcp,langgraph,langsmith,llm-observability","timeRequired":"PT6M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"LangChain","item":"https://daily.dev/sources/langchain"},{"@type":"ListItem","position":3,"name":"How Apollo Uses Deep Agents and LangSmith for GTM AI"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/how-apollo-uses-deep-agents-and-langsmith-for-gtm-ai-2nzctoe3y#faq","mainEntity":[{"@type":"Question","name":"Why did Apollo switch from a LangGraph supervisor architecture to Deep Agents for its AI assistant?","acceptedAnswer":{"@type":"Answer","text":"Apollo moved from a LangGraph supervisor-based multi-agent system to LangChain's Deep Agents because the deterministic graph structure required writing and wiring a new sub-agent for every use case, causing frequent user confirmation prompts. Deep Agents let the assistant dynamically select from a skills library based on stated intent, cutting development time from initial build to launch by roughly 80-85% and reducing confirmation prompts. Engineers weighing agent framework migrations can find real-world architecture writeups like this on daily.dev."}},{"@type":"Question","name":"Why did Apollo choose Deep Agents over Anthropic's Claude Agent SDK for its GTM AI assistant?","acceptedAnswer":{"@type":"Answer","text":"Apollo chose Deep Agents over the Claude Agent SDK because Claude SDK would have locked the assistant into Anthropic models only, while Deep Agents preserves model neutrality across multiple LLM vendors. Apollo's engineering team viewed vendor lock-in as closing doors permanently, making model flexibility the deciding factor. Teams comparing agent SDKs for vendor lock-in tradeoffs follow these evaluations on daily.dev."}},{"@type":"Question","name":"What is Apollo's AI Watchtower evaluation framework for AI agents?","acceptedAnswer":{"@type":"Answer","text":"AI Watchtower is Apollo's six-layer evaluation strategy built on LangSmith, covering pre-launch quality rubrics scored by two reviewers, end-to-end golden-scenario tests run on every PR, live trace debugging, aggregated live insights with drill-downs on quality drops, a weekly pulse report by cohort, and customer thumbs-up/down feedback that triggers a P1 incident if thumbs-down exceeds 8% over a rolling 7-day window. Anyone designing eval pipelines for production AI agents can track similar approaches on daily.dev."}}]} ```