--- title: How agents can delegate better | daily.dev description: Google Cloud and Google DeepMind outline four principles for building AI agents that delegate tasks well within multi-agent systems, drawing on the research... canonical: https://daily.dev/posts/how-agents-can-delegate-better-urftarrex twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: How agents can delegate better | daily.dev og:description: Google Cloud and Google DeepMind outline four principles for building AI agents that delegate tasks well within multi-agent systems, drawing on the research... og:url: https://daily.dev/posts/how-agents-can-delegate-better-urftarrex og:image: https://api.daily.dev/og/posts/urfTaRRex.png og:image:alt: How agents can delegate better 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 agents can delegate better **[Google Cloud](https://daily.dev/sources/gcp)** · 6 min read · 2 upvotes · 0 comments ## Summary Google Cloud and Google DeepMind outline four principles for building AI agents that delegate tasks well within multi-agent systems, drawing on the research paper 'Intelligent AI Delegation'. The principles cover contract-first task decomposition for verifiable work, cost-aware model routing between cheap and powerful models, minimal-permission data sharing backed by zero-knowledge proofs for privacy-preserving verification, and avoiding an overly broad 'zone of indifference' where agents blindly comply with ambiguous requests, instead applying dynamic cognitive friction to flag issues or request human oversight. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What are best practices for building multi-agent AI systems that delegate tasks safely? Four principles help: verify delegated work through contract-first decomposition so tasks can be reliably graded; route tasks to the cheapest model capable of the job rather than always using a strong reasoning model; limit data shared with sub-agents to the minimum needed, using techniques like zero-knowledge proofs to verify work without exposing sensitive data; and avoid a broad 'zone of indifference' where agents blindly comply with ambiguous requests instead of flagging them for human review. _daily.dev surfaces research and engineering practices for teams designing safer multi-agent delegation._ ### How can zero-knowledge proofs be used in AI agent delegation to protect sensitive data? An agent analyzing sensitive data, such as payroll information, can generate a succinct non-interactive argument of knowledge (a type of zero-knowledge proof) that proves a specific property of its computation's result without revealing the underlying data. This lets a delegating agent instantly verify the work was done correctly while the sub-agent's context window stays free of unnecessary sensitive information. _Developers weighing privacy-preserving verification methods track cryptography techniques like this on daily.dev._ ## Similar posts on daily.dev - [Schneier on Security](https://daily.dev/posts/schneier-on-security-xvyc5jov6) · Schneier on Security · 0 upvotes · 0 comments - [Google Explores Scaling Principles for Multi-agent Coordination](https://daily.dev/posts/google-explores-scaling-principles-for-multi-agent-coordination-cqjztktjp) · InfoQ · 0 upvotes · 0 comments - [From Using AI to Delegating Work to AI](https://daily.dev/posts/from-using-ai-to-delegating-work-to-ai-uusgxy83n) · Fatbobman · 0 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#gcp](https://daily.dev/tags/gcp), [#ai-gateway](https://daily.dev/tags/ai-gateway), [#zero-knowledge-proofs](https://daily.dev/tags/zero-knowledge-proofs) [View this post on daily.dev](https://daily.dev/posts/how-agents-can-delegate-better-urftarrex) ```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 agents can delegate better","url":"https://daily.dev/posts/how-agents-can-delegate-better-urftarrex","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/how-agents-can-delegate-better-urftarrex"},"datePublished":"2026-08-21T16:07:22.698Z","dateModified":"2026-09-14T07:06:31.391Z","description":"Google Cloud and Google DeepMind outline four principles for building AI agents that delegate tasks well within multi-agent systems, drawing on the research...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/21522d9a697e58379333cbda3ff8ce6d?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/21522d9a697e58379333cbda3ff8ce6d?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Google Cloud","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":"Google Cloud","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/gcp","url":"https://daily.dev/sources/gcp"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/how-agents-can-delegate-better-urftarrex","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":2},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,gcp,ai-gateway,zero-knowledge-proofs","timeRequired":"PT6M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Google Cloud","item":"https://daily.dev/sources/gcp"},{"@type":"ListItem","position":3,"name":"How agents can delegate better"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/how-agents-can-delegate-better-urftarrex#faq","mainEntity":[{"@type":"Question","name":"What are best practices for building multi-agent AI systems that delegate tasks safely?","acceptedAnswer":{"@type":"Answer","text":"Four principles help: verify delegated work through contract-first decomposition so tasks can be reliably graded; route tasks to the cheapest model capable of the job rather than always using a strong reasoning model; limit data shared with sub-agents to the minimum needed, using techniques like zero-knowledge proofs to verify work without exposing sensitive data; and avoid a broad 'zone of indifference' where agents blindly comply with ambiguous requests instead of flagging them for human review. daily.dev surfaces research and engineering practices for teams designing safer multi-agent delegation."}},{"@type":"Question","name":"How can zero-knowledge proofs be used in AI agent delegation to protect sensitive data?","acceptedAnswer":{"@type":"Answer","text":"An agent analyzing sensitive data, such as payroll information, can generate a succinct non-interactive argument of knowledge (a type of zero-knowledge proof) that proves a specific property of its computation's result without revealing the underlying data. This lets a delegating agent instantly verify the work was done correctly while the sub-agent's context window stays free of unnecessary sensitive information. Developers weighing privacy-preserving verification methods track cryptography techniques like this on daily.dev."}}]} ```