--- title: Pipelines that build themselves | daily.dev description: A pattern called runtime-topology pipelines lets an intake reasoner read an input, decide which analysis dimensions apply, and spawn exactly those specialist... canonical: https://daily.dev/posts/pipelines-that-build-themselves-xxrp9avx6 twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Pipelines that build themselves | daily.dev og:description: A pattern called runtime-topology pipelines lets an intake reasoner read an input, decide which analysis dimensions apply, and spawn exactly those specialist... og:url: https://daily.dev/posts/pipelines-that-build-themselves-xxrp9avx6 og:image: https://api.daily.dev/og/posts/XXRp9AVX6.png og:image:alt: Pipelines that build themselves 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. # Pipelines that build themselves **[AgentField.ai](https://daily.dev/sources/agentfield)** · 8 min read · 0 upvotes · 0 comments ## Summary A pattern called runtime-topology pipelines lets an intake reasoner read an input, decide which analysis dimensions apply, and spawn exactly those specialist workers with prompts written at call time, rather than following a pre-declared graph. Using the AgentField framework, a classifier returns a structured plan, the orchestrator writes runtime prompts and dispatches analysts in parallel via asyncio.gather, a budget counter caps follow-up deep dives, and a synthesis step merges whatever results came back. The underlying control plane handles routing, queueing, tracing, and retries across distributed workers. Example code covers an entry reasoner, analyst reasoner, and curl calls to run and poll the pipeline asynchronously. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How do I build an AI pipeline where the steps that run depend on the input instead of a fixed graph? Use a runtime-topology pattern: an intake reasoner classifies the input and returns a structured plan naming only the analysis dimensions that apply, then the orchestrator writes a specific prompt for each planned dimension and dispatches a generic analyst worker per dimension in parallel with asyncio.gather. The resulting DAG is a trace of what happened, not a graph declared upfront, so a short document might run one dimension while a dense one runs five. _Developers designing adaptive agent workflows can find implementation patterns like this one on daily.dev._ ### How do you stop an AI agent from recursively going deeper and deeper on follow-up analysis without any limit? Cap follow-ups with an explicit budget counter passed into each call, not by relying on the model's judgment. A parent dispatch sets budget=1, allowing exactly one deeper pass; that child pass is then dispatched with budget=0, so it cannot spawn another follow-up. Without this hard counter, an adaptive pipeline tends to find a reason to go one level deeper on every edge case, with no ceiling on cost. _Engineers controlling AI agent costs can track patterns like budget caps on daily.dev._ ## Similar posts on daily.dev - [Build agents, not pipelines](https://daily.dev/posts/build-agents-not-pipelines-2xexg1rzl) · sean goedecke · 6 upvotes · 1 comments - [Building an Advanced Agentic Harness](https://daily.dev/posts/building-an-advanced-agentic-harness-oia5opvy8) · Hacker News · 0 upvotes · 0 comments - [How Grab is Using AI Agents to Boost Team Productivity](https://daily.dev/posts/how-grab-is-using-ai-agents-to-boost-team-productivity-fkppsh0tc) · ByteByteGo · 2 upvotes · 0 comments --- Tags: [#python](https://daily.dev/tags/python), [#llm](https://daily.dev/tags/llm), [#ai-agents](https://daily.dev/tags/ai-agents), [#orchestration](https://daily.dev/tags/orchestration) [View this post on daily.dev](https://daily.dev/posts/pipelines-that-build-themselves-xxrp9avx6) ```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":"Pipelines that build themselves","url":"https://daily.dev/posts/pipelines-that-build-themselves-xxrp9avx6","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/pipelines-that-build-themselves-xxrp9avx6"},"datePublished":"2026-10-02T17:45:21.595Z","dateModified":"2026-10-03T17:36:37.892Z","description":"A pattern called runtime-topology pipelines lets an intake reasoner read an input, decide which analysis dimensions apply, and spawn exactly those specialist...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/1ad566779c58d6672214421e0408ca96?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/1ad566779c58d6672214421e0408ca96?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"AgentField.ai","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":"AgentField.ai","logo":"https://media.daily.dev/image/upload/s--VLFHaedl--/f_auto/v1791063152/squads/33beeb91-1f14-4d9b-9cf8-db7d9f69160a?_a=BAMAMicg0","url":"https://daily.dev/squads/agentfield"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/pipelines-that-build-themselves-xxrp9avx6","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"python,llm,ai-agents,orchestration","timeRequired":"PT8M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"AgentField.ai","item":"https://daily.dev/squads/agentfield"},{"@type":"ListItem","position":3,"name":"Pipelines that build themselves"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/pipelines-that-build-themselves-xxrp9avx6#faq","mainEntity":[{"@type":"Question","name":"How do I build an AI pipeline where the steps that run depend on the input instead of a fixed graph?","acceptedAnswer":{"@type":"Answer","text":"Use a runtime-topology pattern: an intake reasoner classifies the input and returns a structured plan naming only the analysis dimensions that apply, then the orchestrator writes a specific prompt for each planned dimension and dispatches a generic analyst worker per dimension in parallel with asyncio.gather. The resulting DAG is a trace of what happened, not a graph declared upfront, so a short document might run one dimension while a dense one runs five. Developers designing adaptive agent workflows can find implementation patterns like this one on daily.dev."}},{"@type":"Question","name":"How do you stop an AI agent from recursively going deeper and deeper on follow-up analysis without any limit?","acceptedAnswer":{"@type":"Answer","text":"Cap follow-ups with an explicit budget counter passed into each call, not by relying on the model's judgment. A parent dispatch sets budget=1, allowing exactly one deeper pass; that child pass is then dispatched with budget=0, so it cannot spawn another follow-up. Without this hard counter, an adaptive pipeline tends to find a reason to go one level deeper on every edge case, with no ceiling on cost. Engineers controlling AI agent costs can track patterns like budget caps on daily.dev."}}]} ```