<!-- mobian-agent-page publisher="dailydev" canonical="https://daily.dev/posts/three-sub-agent-patterns-you-need-for-your-agentic-system-hvunpjois" -->

---
title: Three sub-agent patterns you need for your agentic system
description: Every production agentic system ends up needing three sub-agent delegation patterns: sync (block and wait), async (fire and forget, reply directly to user),...
canonical: https://daily.dev/posts/three-sub-agent-patterns-you-need-for-your-agentic-system-hvunpjois
twitter:card: summary_large_image
twitter:site: @dailydotdev
og:type: website
og:site_name: daily.dev
og:title: Three sub-agent patterns you need for your agentic system | daily.dev
og:description: Every production agentic system ends up needing three sub-agent delegation patterns: sync (block and wait), async (fire and forget, reply directly to user),...
og:url: https://daily.dev/posts/three-sub-agent-patterns-you-need-for-your-agentic-system-hvunpjois
og:image: https://api.daily.dev/og/posts/HVuNPJOIs.png
og:image:alt: Three sub-agent patterns you need for your agentic system
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.

# Three sub-agent patterns you need for your agentic system

**[Inngest Blog](https://daily.dev/sources/inngest)** · 14 min read · 0 upvotes · 0 comments

## Summary

Every production agentic system ends up needing three sub-agent delegation patterns: sync (block and wait), async (fire and forget, reply directly to user), and scheduled (run later with fresh data). Using Inngest's step.invoke() and step.sendEvent() primitives, one reusable function handles all three modes, differentiated only by trigger and framing instructions. Sub-agents primarily deliver context compression (90%+ token reduction) rather than parallelism. The piece also argues generic sub-agents now beat specialized multi-agent architectures with custom routers, citing Anthropic's 'Building Effective Agents' and the Cursor team's approach, while noting specialization still wins for different model needs, security boundaries, regulatory isolation, or proven eval-backed gains.

## Full article

daily.dev links to this article rather than hosting it. Read it at the original source: <https://www.inngest.com/blog/three-patterns-you-need-for-agentic-systems>

## Questions this post answers

### What are the three core patterns for delegating work to sub-agents in an agentic system?

The three patterns are sync, async, and scheduled delegation. Sync blocks the parent agent until the sub-agent returns a result, used when the next step depends on the answer. Async fires off a sub-agent that replies directly to the user without blocking the parent, and should be the default choice. Scheduled delegates a task to run at a specific future time with fresh data at execution time.

_Anyone architecting agent delegation logic can compare these tradeoffs alongside other implementations on daily.dev._

### Why do sub-agents in an agentic system mainly help with context management rather than parallelism?

Sub-agents reduce the tokens added to a parent agent's context window by over 90% in testing, because the sub-agent runs in its own isolated context, reads files and makes many tool calls, and returns only a compact summary rather than the full trajectory. That context compression, not parallel execution, is described as the real return on investment for using sub-agents.

_Developers weighing context limits versus parallelism in agent design can track this reasoning on daily.dev._

### Should I build specialized sub-agents for different domains or one generic sub-agent for my agentic system?

A single generic sub-agent is recommended over specialized domain agents, because specialization introduces a routing layer that can misroute tasks, creates an eval surface that grows with N agents plus routing and pairwise interaction tests, and often adds complexity without measurable benefit as general-purpose models improve. Specialization is only justified by different model requirements, security boundaries, regulatory needs, or proven eval-driven evidence.

_Teams debating generic versus specialized agent architecture can weigh this tradeoff using resources on daily.dev._

## Similar posts on daily.dev

- [Choosing the Right Multi-Agent Architecture](https://daily.dev/posts/choosing-the-right-multi-agent-architecture-qb5ngrz3j) · LangChain · 2 upvotes · 0 comments
- [Multi-agent coordination patterns: Five approaches and when to use them](https://daily.dev/posts/multi-agent-coordination-patterns-five-approaches-and-when-to-use-them-sywrpvaao) · Claude · 1 upvotes · 0 comments
- [Claude Subagents vs. Agent Teams](https://daily.dev/posts/claude-subagents-vs-agent-teams-2nvftotxu) · Daily Dose of Data Science \| Avi Chawla \| Substack · 3 upvotes · 1 comments
- [4 engineering patterns behind the strongest AI Agents Challenge submissions](https://daily.dev/posts/4-engineering-patterns-behind-the-strongest-ai-agents-challenge-submissions-v5ze5ppnk) · Google Developers · 0 upvotes · 0 comments

---

Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#inngest](https://daily.dev/tags/inngest)

[View this post on daily.dev](https://daily.dev/posts/three-sub-agent-patterns-you-need-for-your-agentic-system-hvunpjois)

```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":"Three sub-agent patterns you need for your agentic system","url":"https://daily.dev/posts/three-sub-agent-patterns-you-need-for-your-agentic-system-hvunpjois","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/three-sub-agent-patterns-you-need-for-your-agentic-system-hvunpjois"},"datePublished":"2026-09-13T19:57:34.053Z","dateModified":"2026-10-15T02:22:39.676Z","description":"Every production agentic system ends up needing three sub-agent delegation patterns: sync (block and wait), async (fire and forget, reply directly to user),...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/ce50a63c85506cf895f69f9338572b7b?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/ce50a63c85506cf895f69f9338572b7b?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Inngest Blog","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":"Inngest Blog","logo":"https://media.daily.dev/image/upload/s--ieZKUa5y--/c_limit,w_256/f_auto,q_auto/v1789329423/logos/inngest?_a=BAMAMicg0","url":"https://daily.dev/sources/inngest"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/three-sub-agent-patterns-you-need-for-your-agentic-system-hvunpjois","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,inngest","timeRequired":"PT14M"}
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Inngest Blog","item":"https://daily.dev/sources/inngest"},{"@type":"ListItem","position":3,"name":"Three sub-agent patterns you need for your agentic system"}]}
{"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/three-sub-agent-patterns-you-need-for-your-agentic-system-hvunpjois#faq","mainEntity":[{"@type":"Question","name":"What are the three core patterns for delegating work to sub-agents in an agentic system?","acceptedAnswer":{"@type":"Answer","text":"The three patterns are sync, async, and scheduled delegation. Sync blocks the parent agent until the sub-agent returns a result, used when the next step depends on the answer. Async fires off a sub-agent that replies directly to the user without blocking the parent, and should be the default choice. Scheduled delegates a task to run at a specific future time with fresh data at execution time. Anyone architecting agent delegation logic can compare these tradeoffs alongside other implementations on daily.dev."}},{"@type":"Question","name":"Why do sub-agents in an agentic system mainly help with context management rather than parallelism?","acceptedAnswer":{"@type":"Answer","text":"Sub-agents reduce the tokens added to a parent agent's context window by over 90% in testing, because the sub-agent runs in its own isolated context, reads files and makes many tool calls, and returns only a compact summary rather than the full trajectory. That context compression, not parallel execution, is described as the real return on investment for using sub-agents. Developers weighing context limits versus parallelism in agent design can track this reasoning on daily.dev."}},{"@type":"Question","name":"Should I build specialized sub-agents for different domains or one generic sub-agent for my agentic system?","acceptedAnswer":{"@type":"Answer","text":"A single generic sub-agent is recommended over specialized domain agents, because specialization introduces a routing layer that can misroute tasks, creates an eval surface that grows with N agents plus routing and pairwise interaction tests, and often adds complexity without measurable benefit as general-purpose models improve. Specialization is only justified by different model requirements, security boundaries, regulatory needs, or proven eval-driven evidence. Teams debating generic versus specialized agent architecture can weigh this tradeoff using resources on daily.dev."}}]}
```

