--- title: Every Harness Will Become A Claw — Sam Bhagwat, Mastra description: Sam Bhagwat, co-founder of Mastra (a TypeScript agent framework), presents a framework for understanding the evolution of AI agents. He introduces the concept... canonical: https://daily.dev/posts/every-harness-will-become-a-claw-sam-bhagwat-mastra-ii98cciyv twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Every Harness Will Become A Claw — Sam Bhagwat, Mastra | daily.dev og:description: Sam Bhagwat, co-founder of Mastra (a TypeScript agent framework), presents a framework for understanding the evolution of AI agents. He introduces the concept... og:url: https://daily.dev/posts/every-harness-will-become-a-claw-sam-bhagwat-mastra-ii98cciyv og:image: https://api.daily.dev/og/posts/Ii98CcIyv.png og:image:alt: Every Harness Will Become A Claw — Sam Bhagwat, Mastra 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. # Every Harness Will Become A Claw — Sam Bhagwat, Mastra **[AI Engineer](https://daily.dev/sources/aidotengineer)** · 15 min read · 3 upvotes · 0 comments ## Summary Sam Bhagwat, co-founder of Mastra (a TypeScript agent framework), presents a framework for understanding the evolution of AI agents. He introduces the concept of the 'harness era' — distinguishing between local harnesses (like coding assistants), cloud harnesses (always-on, Slack-integrated, parallel sub-agents), and 'claws' (agents with initiative, heartbeats, and continual learning). He proposes 'Steinberger's Law': every harness will expand until it becomes a claw. Drawing a parallel to the mobile app ecosystem of the 2010s, he predicts a coming shakeout where only the most economically valuable or frequently used agent claws will survive in users' mental space. His advice to builders: ensure your agents have the capabilities users need, and stay current as the rate of change accelerates. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is the difference between an AI agent, a harness, and a claw? An agent adds a loop, tool calls, memory, retries, and context engineering on top of a plain LLM call. A harness adds durability and doggedness: it can run for hours or days, persist and resume threads, spin up parallel sub-agents, and autocompact context, as seen in tools like Claude Code or Codex. A claw goes further, adding initiative and continual learning: it listens to external feeds, has a heartbeat that wakes it periodically, operates across channels like Slack, WhatsApp, or Telegram, and improves itself from its own traces. _Track how coding agents evolve from harnesses into always-on claws by following agent framework updates on daily.dev._ ### What is Steinberger's law in AI agent development? Steinberger's law, coined during a Mastra talk, states that every harness will expand until it becomes a claw. The reasoning is partly technological, partly economic, and partly psychological: users want to message agents in Slack, kick off overnight tasks before bed, and repeatedly experience the reward of turning tokens into useful output, which pushes every harness toward more autonomous, always-on behavior. _Developers weighing how much autonomy to build into their agents can follow this debate on daily.dev._ ### Why might most AI agent categories fail to survive long term according to predictions about the agent market? A predicted shakeout mirrors what happened with 2010s mobile apps: despite dozens of app categories emerging, users only kept one or two dominant apps per category, like Uber and Lyft for rides, because attention is limited and a service must be either highly economically valuable or used very frequently to stay relevant. The same dynamic is expected to hit AI agent harnesses, leaving room for only a handful of dominant agent categories per use case. _Anyone choosing which agent tools to bet on can follow how this consolidation unfolds on daily.dev._ ## Similar posts on daily.dev - [The Evolution of the Agent Harness](https://daily.dev/posts/the-evolution-of-the-agent-harness-1z25xrcos) · Latent Space · 8 upvotes · 1 comments - [Meta Harnesses, Agents, and Lessons from the Framework Wars](https://daily.dev/posts/meta-harnesses-agents-and-lessons-from-the-framework-wars-ajel9z5tw) · Nx · 16 upvotes · 1 comments - [Harness Engineering](https://daily.dev/posts/harness-engineering-l1sahcwor) · Martin Fowler · 4 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#ai-agents](https://daily.dev/tags/ai-agents), [#typescript](https://daily.dev/tags/typescript) [View this post on daily.dev](https://daily.dev/posts/every-harness-will-become-a-claw-sam-bhagwat-mastra-ii98cciyv) ```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":"Every Harness Will Become A Claw — Sam Bhagwat, Mastra","url":"https://daily.dev/posts/every-harness-will-become-a-claw-sam-bhagwat-mastra-ii98cciyv","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/every-harness-will-become-a-claw-sam-bhagwat-mastra-ii98cciyv"},"datePublished":"2026-07-21T21:46:14.664Z","dateModified":"2026-09-13T21:31:21.686Z","description":"Sam Bhagwat, co-founder of Mastra (a TypeScript agent framework), presents a framework for understanding the evolution of AI agents. He introduces the concept...","image":"https://i.ytimg.com/vi/8qWIPUia2O8/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/8qWIPUia2O8/sddefault.jpg","isAccessibleForFree":true,"articleSection":"AI Engineer","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":"AI Engineer","logo":"https://media.daily.dev/image/upload/s--u5PucxNT--/f_auto/v1724338940/logos/aidotengineer","url":"https://daily.dev/sources/aidotengineer"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/every-harness-will-become-a-claw-sam-bhagwat-mastra-ii98cciyv","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":3},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,ai-agents,typescript","timeRequired":"PT15M","video":{"@type":"VideoObject","name":"Every Harness Will Become A Claw — Sam Bhagwat, Mastra","description":"Sam Bhagwat, co-founder of Mastra (a TypeScript agent framework), presents a framework for understanding the evolution of AI agents. He introduces the concept...","thumbnailUrl":"https://i.ytimg.com/vi/8qWIPUia2O8/sddefault.jpg","uploadDate":"2026-07-21T21:46:14.664Z","duration":"PT15M","url":"https://api.daily.dev/r/Ii98CcIyv","embedUrl":"https://www.youtube.com/embed/8qWIPUia2O8"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"AI Engineer","item":"https://daily.dev/sources/aidotengineer"},{"@type":"ListItem","position":3,"name":"Every Harness Will Become A Claw — Sam Bhagwat, Mastra"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/every-harness-will-become-a-claw-sam-bhagwat-mastra-ii98cciyv#faq","mainEntity":[{"@type":"Question","name":"What is the difference between an AI agent, a harness, and a claw?","acceptedAnswer":{"@type":"Answer","text":"An agent adds a loop, tool calls, memory, retries, and context engineering on top of a plain LLM call. A harness adds durability and doggedness: it can run for hours or days, persist and resume threads, spin up parallel sub-agents, and autocompact context, as seen in tools like Claude Code or Codex. A claw goes further, adding initiative and continual learning: it listens to external feeds, has a heartbeat that wakes it periodically, operates across channels like Slack, WhatsApp, or Telegram, and improves itself from its own traces. Track how coding agents evolve from harnesses into always-on claws by following agent framework updates on daily.dev."}},{"@type":"Question","name":"What is Steinberger's law in AI agent development?","acceptedAnswer":{"@type":"Answer","text":"Steinberger's law, coined during a Mastra talk, states that every harness will expand until it becomes a claw. The reasoning is partly technological, partly economic, and partly psychological: users want to message agents in Slack, kick off overnight tasks before bed, and repeatedly experience the reward of turning tokens into useful output, which pushes every harness toward more autonomous, always-on behavior. Developers weighing how much autonomy to build into their agents can follow this debate on daily.dev."}},{"@type":"Question","name":"Why might most AI agent categories fail to survive long term according to predictions about the agent market?","acceptedAnswer":{"@type":"Answer","text":"A predicted shakeout mirrors what happened with 2010s mobile apps: despite dozens of app categories emerging, users only kept one or two dominant apps per category, like Uber and Lyft for rides, because attention is limited and a service must be either highly economically valuable or used very frequently to stay relevant. The same dynamic is expected to hit AI agent harnesses, leaving room for only a handful of dominant agent categories per use case. Anyone choosing which agent tools to bet on can follow how this consolidation unfolds on daily.dev."}}]} ```