--- title: Understanding Human-Agent Collaboration Patterns | daily.dev description: Human-agent collaboration goes beyond keeping a person in the loop. It defines four delegation patterns: assistance, approval before action, action with... canonical: https://daily.dev/posts/understanding-human-agent-collaboration-patterns-fbkpv3dki twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Understanding Human-Agent Collaboration Patterns | daily.dev og:description: Human-agent collaboration goes beyond keeping a person in the loop. It defines four delegation patterns: assistance, approval before action, action with... og:url: https://daily.dev/posts/understanding-human-agent-collaboration-patterns-fbkpv3dki og:image: https://api.daily.dev/og/posts/fbkpv3dki.png og:image:alt: Understanding Human-Agent Collaboration Patterns 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. # Understanding Human-Agent Collaboration Patterns **[Telerik](https://daily.dev/sources/telerik)** · 8 min read · 0 upvotes · 0 comments ## Summary Human-agent collaboration goes beyond keeping a person in the loop. It defines four delegation patterns: assistance, approval before action, action with review, and bounded delegation, each suited to different levels of risk. Approval halts an action before execution for consequential or hard-to-reverse cases, while review checks completed work after the fact for reversible tasks. Escalation transfers responsibility to a person when an agent hits a policy limit, conflicting data, or validation failure, and needs an owner and response target. A TypeScript example shows application code, not the model, enforcing refund limits and triggering approval or escalation. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### When should an AI agent require approval before acting versus just a review afterward? Approval works best for big, costly, or hard-to-reverse actions, or ones that create an external commitment or modify a customer's access; the reviewer should see the action and supporting evidence before it executes. Review works best for easily reversible actions where the system can detect common failures, letting a team check completed work or a sample afterward without blocking every run. _Teams designing approval gates for agents can track emerging patterns like this on daily.dev._ ### What triggers should cause an AI agent to escalate a task to a human instead of just failing? Reliable escalation triggers include a task exceeding the agent's defined scope or financial limit, failure to validate required data, or two data sources giving contradictory instructions. These rule-based triggers are more dependable than asking an agent to judge its own confidence, and escalation needs an assigned owner and response target so paused tasks don't sit unnoticed. _Developers building agent workflows weigh escalation rules like these on daily.dev._ ### Where should the controls that limit an AI agent's actions actually live in an application? The strongest controls live outside the model, in application code: tool permissions constrain what the agent can access, schema validation limits the arguments it sends, policy checks decide when work can continue, and audit records preserve what happened. The model proposes actions, like a refund request, but application code validates and applies the limits the business sets. _Developers architecting agent guardrails compare approaches like this on daily.dev._ ## Similar posts on daily.dev - [One Agent Workflow That Keeps Human Review in the Loop](https://daily.dev/posts/one-agent-workflow-that-keeps-human-review-in-the-loop-idljoj6z9) · Medium · 0 upvotes · 0 comments - [Top AI Agentic Workflow Patterns](https://daily.dev/posts/top-ai-agentic-workflow-patterns-kh07ttxry) · ByteByteGo · 9 upvotes · 0 comments - [8 Agentic AI patterns reshaping team collaboration](https://daily.dev/posts/8-agentic-ai-patterns-reshaping-team-collaboration-1kv1ojet8) · GitLab · 27 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#career](https://daily.dev/tags/career), [#typescript](https://daily.dev/tags/typescript), [#ai-safety](https://daily.dev/tags/ai-safety) [View this post on daily.dev](https://daily.dev/posts/understanding-human-agent-collaboration-patterns-fbkpv3dki) ```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":"Understanding Human-Agent Collaboration Patterns","url":"https://daily.dev/posts/understanding-human-agent-collaboration-patterns-fbkpv3dki","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/understanding-human-agent-collaboration-patterns-fbkpv3dki"},"datePublished":"2026-10-28T14:12:10.417Z","dateModified":"2026-10-28T15:00:35.117Z","description":"Human-agent collaboration goes beyond keeping a person in the loop. It defines four delegation patterns: assistance, approval before action, action with...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/b83e6573037525bc78570b48599b4cd0?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/b83e6573037525bc78570b48599b4cd0?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Telerik","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":"Telerik","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/5ee6d4bef8ac4f84886d06f5881963e7","url":"https://daily.dev/sources/telerik"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/understanding-human-agent-collaboration-patterns-fbkpv3dki","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,career,typescript,ai-safety","timeRequired":"PT8M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Telerik","item":"https://daily.dev/sources/telerik"},{"@type":"ListItem","position":3,"name":"Understanding Human-Agent Collaboration Patterns"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/understanding-human-agent-collaboration-patterns-fbkpv3dki#faq","mainEntity":[{"@type":"Question","name":"When should an AI agent require approval before acting versus just a review afterward?","acceptedAnswer":{"@type":"Answer","text":"Approval works best for big, costly, or hard-to-reverse actions, or ones that create an external commitment or modify a customer's access; the reviewer should see the action and supporting evidence before it executes. Review works best for easily reversible actions where the system can detect common failures, letting a team check completed work or a sample afterward without blocking every run. Teams designing approval gates for agents can track emerging patterns like this on daily.dev."}},{"@type":"Question","name":"What triggers should cause an AI agent to escalate a task to a human instead of just failing?","acceptedAnswer":{"@type":"Answer","text":"Reliable escalation triggers include a task exceeding the agent's defined scope or financial limit, failure to validate required data, or two data sources giving contradictory instructions. These rule-based triggers are more dependable than asking an agent to judge its own confidence, and escalation needs an assigned owner and response target so paused tasks don't sit unnoticed. Developers building agent workflows weigh escalation rules like these on daily.dev."}},{"@type":"Question","name":"Where should the controls that limit an AI agent's actions actually live in an application?","acceptedAnswer":{"@type":"Answer","text":"The strongest controls live outside the model, in application code: tool permissions constrain what the agent can access, schema validation limits the arguments it sends, policy checks decide when work can continue, and audit records preserve what happened. The model proposes actions, like a refund request, but application code validates and applies the limits the business sets. Developers architecting agent guardrails compare approaches like this on daily.dev."}}]} ```