--- title: OpenAI reportedly planning Managed Agents announcement... description: OpenAI is rumored to unveil a Managed Agents feature at its DevDay event on September 29, based on leaked details from a social media account. The feature... canonical: https://daily.dev/posts/openai-reportedly-planning-managed-agents-announcement-at-devday-2026-idgeupswk twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: OpenAI reportedly planning Managed Agents announcement at DevDay 2026 | daily.dev og:description: OpenAI is rumored to unveil a Managed Agents feature at its DevDay event on September 29, based on leaked details from a social media account. The feature... og:url: https://daily.dev/posts/openai-reportedly-planning-managed-agents-announcement-at-devday-2026-idgeupswk og:image: https://api.daily.dev/og/posts/iDGEuPSwK.png og:image:alt: OpenAI reportedly planning Managed Agents announcement at DevDay 2026 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. # OpenAI reportedly planning Managed Agents announcement at DevDay 2026 **[Collections](https://daily.dev/sources/collections)** · 4 min read · 0 upvotes · 0 comments ## Summary OpenAI is rumored to unveil a Managed Agents feature at its DevDay event on September 29, based on leaked details from a social media account. The feature would reportedly let developers create Agents, Environments, and Agent Sessions directly through the OpenAI Platform, alongside a Codex Plugin for building and deploying agents with minimal manual setup. Hosting could be flexible, either on OpenAI's platform or self-hosted. A developer hinted at upcoming related work, but OpenAI has not confirmed anything officially. ## Content OpenAI has launched a managed Agents API in public beta, and it's worth understanding what's actually new here before the hype cycle takes over. ## What it does The Agents API is essentially the harness behind Codex, now exposed as an API. You give it a task, a model, a set of tools, and a sandbox environment. OpenAI handles the rest: the agent loop, orchestration, long-running sessions, context management, and retries. The infrastructure work that used to eat weeks of engineering time - job queues, state databases, retry policies - is now abstracted away. Agents can run in an OpenAI-managed sandbox, your own infrastructure, or third-party providers including Cloudflare, DigitalOcean, and Vercel. Box has also announced a filesystem integration, letting agents read and write files directly to enterprise storage during execution. OpenAI is also planning a Codex Plugin that lets users build, configure, and deploy agents through a conversational interface - lowering the barrier further for teams without dedicated ML infrastructure. ## Use cases being targeted OpenAI has outlined several reference agent types: - **Research agents** that search the web and connected data sources to produce structured output - **Customer support agents** that retrieve account data and escalate when needed - **DevOps assistants** that monitor deployments and help troubleshoot incidents - **Meeting assistants** that transcribe, capture action items, and summarize - **Analytics agents** that query data sources and explain trends in plain language ## Why the open-source harness matters One underappreciated angle: the Codex harness is open-source. That means other models - open and closed - are already trained on it and understand its conventions. As more services adopt this architecture, there's a compounding effect from that shared knowledge. A harness-as-a-service built on an open standard is a different proposition than one built on proprietary scaffolding. ## The competitive landscape OpenAI isn't alone here. Over the past few months, nearly every major AI company has shipped something in this space: - Anthropic's Claude Managed Agents (April) - Google's Antigravity and Gemini agents (May) - AWS AgentCore (June) - Microsoft Foundry hosted agents (June) - LangChain's managed Deep Agents beta (August) Each has a different lock-in strategy. Model companies like Anthropic lock you into their models. Cloud providers like AWS make models interchangeable but lock you into their runtime. LangChain's play is locking you in through LangSmith observability tooling. OpenAI's approach sits somewhere between model and runtime lock-in. ## The real concerns The engineering simplification is genuine. But a few things are worth thinking through before committing: **No Zero Data Retention support.** Session and sandbox state stay server-side. For regulated industries - healthcare, finance, legal - this could be a dealbreaker. **Negotiating leverage.** Once your agent architecture is built around OpenAI's runtime, switching costs go up significantly. That weakens your position in future pricing conversations. **Billing structure.** Managed agent products typically layer session-hour billing on top of token costs. A single complex interaction can burn 100k to 3M tokens. The economics look different at scale than they do in a prototype. Five questions worth asking before choosing managed agents over self-hosting: 1. Do you need persistent state across sessions? 2. Do you need a server-side sandbox? 3. Do you need full control over the agent loop? 4. Do you have Zero Data Retention requirements? 5. How much does vendor leverage matter to your organization long-term? If the answer to questions 3 or 4 is yes, self-hosting a harness is probably the right call. If you're optimizing for speed to production and the compliance constraints don't apply, the managed path is genuinely faster. The Agents API is a real step forward for teams that want to ship agents without building infrastructure. The tradeoffs are real too, and they're worth mapping out before you're too deep to change course. ## Questions this post answers ### What is OpenAI's Managed Agents feature that's rumored for DevDay 2026? Managed Agents is a reported OpenAI Platform feature letting users create agents, environments, and agent sessions without writing infrastructure code. A Codex plugin would handle build, configure, and deploy, with agents hosted on OpenAI's platform or self-hosted. Rumored use cases include research agents, customer support agents, DevOps assistants, meeting assistants, and analytics agents. It's expected to be announced at DevDay on September 29th, though details come from testing sources rather than an official announcement. _daily.dev keeps developers evaluating agent platforms ahead of announcements like this one._ ## Community take How the wider developer community reacted, aggregated from 5 discussions and 32 comments across x (as of 2026-09-14). **TL;DR:** Reaction centers on the new Box-mounted sandbox and managed agent loop, with excitement about simpler primitives tempered by concern over permissions, auditability, and blast radius when agents get file write access. **Sentiment:** 45% positive · 40% mixed · 15% skeptical **The case for** - The managed agent loop removes the need to hand-wire orchestration. - Mounting a filesystem like Box directly into the sandbox is seen as a simple, powerful primitive that saves time explaining context to agents. - Giving agents the same file primitives humans use is viewed as a natural, overdue step. **The pushback** - Write access into enterprise file systems increases the blast radius when an agent errs. - Permission scoping and audit trails for agent writes are seen as unresolved and likely to cause problems. - Auth and permissions, not read/write itself, are expected to be where agents actually stall. **By community** - x (mixed): Enthusiasm for the new managed agent loop and file-mount primitive is mixed with repeated concern about permissions, write access, and audit trails. **Hottest debate:** Whether mounting enterprise files into agent sandboxes is a clean usability win or a serious risk once agents can write back with real permissions. **Open questions** - How will permission scoping work when an agent writes back into a mounted enterprise filesystem like Box? - Will agent-driven edits get proper audit trails and version history, or will oversight lag behind adoption? **Highlights** > @levie Mounting enterprise files directly into the sandbox is powerful, but it also increases the blast radius when the agent gets something wrong. Read/write access should probably be treated like production access: minimum permissions, hard boundaries and independent checks before > — [\_RanjanSoni on x](https://x.com/_RanjanSoni/status/2098481498706059428) > @levie The primitive people got with file access was an audit trail, not just read/write. Agents doing 100X the edits means 100X the writes nobody reviewed. Mount is the easy half > — [jschoen\_deroolo on x](https://x.com/jschoen_deroolo/status/2098518953287135375) > @levie the primitives point is right but file i/o is still the boring part. agents will stall on auth and permissions long before they stall on read/write. > — [abenz95 on x · 1 points](https://x.com/abenz95/status/2098498498497032257) > @levie This is a massive step for usability, but it's going to put Enterprise Security layers to the test. If an agent can seamlessly read and write to Box, our context-window access control and prompt-injection defense mechanisms need to be absolutely flawless > — [sciphi\_ai on x](https://x.com/sciphi_ai/status/2098511039579246680) > @levie Give an agent a mounted drive and it inherits everything, including the folder where someone saved Final_v3_FINAL_actual.xlsx. The primitives humans had were mostly bad habits. > — [MarcusSpillane on x](https://x.com/MarcusSpillane/status/2098485832504004629) **Source threads** - [x](https://x.com/trikcode/status/2098312807124971580) · 0 points · 8 comments - [x](https://x.com/levie/status/2098478938003841123) · 0 points · 24 comments - [x](https://x.com/OpenAIDevs/status/2098496675216306344) · 0 points · 0 comments - [x](https://x.com/OpenAIDevs/status/2098496733747839147) · 0 points · 0 comments - [x](https://x.com/omarsar0/status/2098524621439914375) · 0 points · 0 comments ## Similar posts on daily.dev - [Anthropic, OpenAI, Google, and Microsoft agree that the harness is the product. They disagree on the price.](https://daily.dev/posts/anthropic-openai-google-and-microsoft-agree-that-the-harness-is-the-product-they-disagree-on-the-hvk3m7gts) · The New Stack · 1 upvotes · 0 comments - [Anthropic Introduces Managed Agents to Simplify AI Agent Deployment](https://daily.dev/posts/anthropic-introduces-managed-agents-to-simplify-ai-agent-deployment-em1vdc2kw) · InfoQ · 1 upvotes · 0 comments - [Harness engineering: leveraging Codex in an agent-first world](https://daily.dev/posts/harness-engineering-leveraging-codex-in-an-agent-first-world-py6m8jwm4) · OpenAI · 1 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#openai](https://daily.dev/tags/openai), [#openai-codex](https://daily.dev/tags/openai-codex) [View this post on daily.dev](https://daily.dev/posts/openai-reportedly-planning-managed-agents-announcement-at-devday-2026-idgeupswk) ```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":"OpenAI reportedly planning Managed Agents announcement at DevDay 2026","url":"https://daily.dev/posts/openai-reportedly-planning-managed-agents-announcement-at-devday-2026-idgeupswk","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/openai-reportedly-planning-managed-agents-announcement-at-devday-2026-idgeupswk"},"datePublished":"2026-09-08T15:53:28.998Z","dateModified":"2026-09-14T04:49:03.656Z","description":"OpenAI is rumored to unveil a Managed Agents feature at its DevDay event on September 29, based on leaked details from a social media account. The feature...","isAccessibleForFree":true,"articleSection":"Collections","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":"Collections","logo":"https://media.daily.dev/image/upload/s--fk_6ycEi--/f_auto,q_auto/v1780996001/logos/collections?_a=BAMAMiWQ0","url":"https://daily.dev/sources/collections"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/openai-reportedly-planning-managed-agents-announcement-at-devday-2026-idgeupswk","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,openai,openai-codex","timeRequired":"PT4M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Collections","item":"https://daily.dev/sources/collections"},{"@type":"ListItem","position":3,"name":"OpenAI reportedly planning Managed Agents announcement at DevDay 2026"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/openai-reportedly-planning-managed-agents-announcement-at-devday-2026-idgeupswk#faq","mainEntity":[{"@type":"Question","name":"What is OpenAI's Managed Agents feature that's rumored for DevDay 2026?","acceptedAnswer":{"@type":"Answer","text":"Managed Agents is a reported OpenAI Platform feature letting users create agents, environments, and agent sessions without writing infrastructure code. A Codex plugin would handle build, configure, and deploy, with agents hosted on OpenAI's platform or self-hosted. Rumored use cases include research agents, customer support agents, DevOps assistants, meeting assistants, and analytics agents. It's expected to be announced at DevDay on September 29th, though details come from testing sources rather than an official announcement. daily.dev keeps developers evaluating agent platforms ahead of announcements like this one."}}]} ```