--- title: Laravel AI SDK 1.0: what's new and what breaks | daily.dev description: Laravel AI SDK 1.0 ships with classification via TypeSafe's Jev models, approvable tool calls that pause for human sign-off before sensitive agent actions,... canonical: https://daily.dev/posts/laravel-ai-sdk-1-0-what-s-new-and-what-breaks-obianoco9 twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Laravel AI SDK 1.0: what's new and what breaks | daily.dev og:description: Laravel AI SDK 1.0 ships with classification via TypeSafe's Jev models, approvable tool calls that pause for human sign-off before sensitive agent actions,... og:url: https://daily.dev/posts/laravel-ai-sdk-1-0-what-s-new-and-what-breaks-obianoco9 og:image: https://api.daily.dev/og/posts/oBIANoco9.png og:image:alt: Laravel AI SDK 1.0: what's new and what breaks 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. # Laravel AI SDK 1.0: what's new and what breaks **[Collections](https://daily.dev/sources/collections)** · 3 min read · 1 upvotes · 0 comments ## Summary Laravel AI SDK 1.0 ships with classification via TypeSafe's Jev models, approvable tool calls that pause for human sign-off before sensitive agent actions, per-step agent middleware, lazy-loaded tool search, Vercel Chat and AG-UI streaming support, and provider-side code execution. The release also brings breaking changes to the conversation storage format (requiring a backfill migration), the agent middleware API, token usage field names, and stream protocols. Teams are advised to run the backfill migration before deploying new code and to use the upgrade guide with an AI coding assistant via Laravel Boost. ## Content ## What shipped in 1.0 Laravel AI SDK hit 1.0 after a long stretch at 0.11, and the release is substantial. Here's what's actually new: **Classification via Jev models.** TypeSafe's Jev models are now integrated, giving you fast, typed decisions - think ticket routing or spam detection - at a fraction of what a full LLM call costs. Taylor Otwell flagged this as one of the headline additions. **Approvable tool calls.** Agents can now pause and wait for human sign-off before running sensitive operations like file deletion. This is the kind of guardrail that makes agents actually deployable in production rather than just demos. **Vercel Chat and AG-UI streaming support.** Both protocols are now supported, making it easier to pair the SDK with frontend chat libraries without rolling your own streaming glue. **Per-step agent middleware.** You can now hook into individual steps of an agent run, which opens up logging, validation, and conditional logic that wasn't practical before. **Tool search for lazy-loading.** Rarely used tools can now be loaded on demand rather than registered upfront. Useful if you have a large tool surface and don't want to pay the overhead on every request. **Provider-side code execution.** Code can now run on the provider's infrastructure rather than locally, depending on what the provider supports. **Consolidated conversation storage format.** The way conversations are stored has been unified, which is cleaner but comes with a migration cost (more on that below). --- ## Breaking changes This isn't a drop-in upgrade. The 1.0 release includes breaking changes to: - Conversation storage format (requires a backfill migration before deploying) - Agent middleware API - Token usage field names (`data` usage renamed to `text` usage) - Stream protocols The Laravel team recommends running the upgrade through Laravel Boost with an AI coding assistant to walk through the guide automatically. --- ## How the AI-assisted upgrade actually goes One developer tested the recommended upgrade path live - using Laravel Boost's upgrade skill with Claude Code (Opus) - and the results were mostly good with some friction. The agent handled the migration automatically, including bumping Laravel MCP to its own v1.0 as a prerequisite before touching the main SDK. The one breaking change in the project's codebase (the `data` → `text` usage rename) was caught and applied without manual intervention. That said, it wasn't frictionless. The Boost skill set was outdated and needed updating before the upgrade skill was even available. The initial install failed once. And the resulting git diff had 119 changed files - mostly skill markdown files - which makes reviewing the actual code changes harder than it should be. The conclusion: the automated path works, but you should still read the diff. Whether you trust an agent to handle dependency upgrades like this is a reasonable question to sit with. ## Questions this post answers ### What breaking changes are in Laravel AI SDK 1.0? Laravel AI SDK 1.0 breaks the conversation storage format (requiring a backfill migration), the agent middleware API, token usage field names, and stream protocols. The backfill migration must run before new code goes live to avoid data issues. The team recommends using an AI coding assistant with Laravel Boost to walk through the upgrade guide step by step. _Track breaking changes like these before upgrading a Laravel AI integration on daily.dev._ ### What new agent features does Laravel AI SDK 1.0 add? Laravel AI SDK 1.0 adds approvable tool calls that let agents pause for human sign-off before sensitive actions like file deletion, per-step agent middleware for granular control, and tool search that lazy-loads rarely used tools instead of registering them upfront. It also adds classification via TypeSafe's Jev models for fast, typed decisions like ticket routing at lower cost than a full LLM call. _Developers evaluating agent safety controls can follow releases like this on daily.dev._ ### Does Laravel AI SDK 1.0 support streaming with Vercel Chat? Yes, Laravel AI SDK 1.0 adds support for both Vercel Chat and AG-UI streaming protocols, making it easier to pair the SDK with frontend chat libraries. The release also introduces provider-side code execution and a consolidated conversation storage format. _daily.dev helps developers pairing backend SDKs with frontend chat tooling stay current on protocol support._ ## Similar posts on daily.dev - [Laravel Boost Update Adds Support for the New MCP Protocol](https://daily.dev/posts/laravel-boost-update-adds-support-for-the-new-mcp-protocol-8827svtqm) · Laravel News · 3 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#ai-agents](https://daily.dev/tags/ai-agents), [#php](https://daily.dev/tags/php), [#laravel](https://daily.dev/tags/laravel) [View this post on daily.dev](https://daily.dev/posts/laravel-ai-sdk-1-0-what-s-new-and-what-breaks-obianoco9) ```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":"Laravel AI SDK 1.0: what's new and what breaks","url":"https://daily.dev/posts/laravel-ai-sdk-1-0-what-s-new-and-what-breaks-obianoco9","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/laravel-ai-sdk-1-0-what-s-new-and-what-breaks-obianoco9"},"datePublished":"2026-09-23T15:45:25.749Z","dateModified":"2026-09-24T07:14:36.763Z","description":"Laravel AI SDK 1.0 ships with classification via TypeSafe's Jev models, approvable tool calls that pause for human sign-off before sensitive agent actions,...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/f793f294ca30781905cef7569576700b?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/f793f294ca30781905cef7569576700b?_a=AQAEuop","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/laravel-ai-sdk-1-0-what-s-new-and-what-breaks-obianoco9","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,ai-agents,php,laravel","timeRequired":"PT3M"} {"@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":"Laravel AI SDK 1.0: what's new and what breaks"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/laravel-ai-sdk-1-0-what-s-new-and-what-breaks-obianoco9#faq","mainEntity":[{"@type":"Question","name":"What breaking changes are in Laravel AI SDK 1.0?","acceptedAnswer":{"@type":"Answer","text":"Laravel AI SDK 1.0 breaks the conversation storage format (requiring a backfill migration), the agent middleware API, token usage field names, and stream protocols. The backfill migration must run before new code goes live to avoid data issues. The team recommends using an AI coding assistant with Laravel Boost to walk through the upgrade guide step by step. Track breaking changes like these before upgrading a Laravel AI integration on daily.dev."}},{"@type":"Question","name":"What new agent features does Laravel AI SDK 1.0 add?","acceptedAnswer":{"@type":"Answer","text":"Laravel AI SDK 1.0 adds approvable tool calls that let agents pause for human sign-off before sensitive actions like file deletion, per-step agent middleware for granular control, and tool search that lazy-loads rarely used tools instead of registering them upfront. It also adds classification via TypeSafe's Jev models for fast, typed decisions like ticket routing at lower cost than a full LLM call. Developers evaluating agent safety controls can follow releases like this on daily.dev."}},{"@type":"Question","name":"Does Laravel AI SDK 1.0 support streaming with Vercel Chat?","acceptedAnswer":{"@type":"Answer","text":"Yes, Laravel AI SDK 1.0 adds support for both Vercel Chat and AG-UI streaming protocols, making it easier to pair the SDK with frontend chat libraries. The release also introduces provider-side code execution and a consolidated conversation storage format. daily.dev helps developers pairing backend SDKs with frontend chat tooling stay current on protocol support."}}]} ```