> ## Documentation Index
> Fetch the complete documentation index at: https://learn.playlab.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# New in Playlab: the Assistant, responsible organizations, Insights, and voice
> What shipped in Q3 2026: the Playlab Assistant, responsible organizations, a redesigned Insights, natural-sounding voice output, new models, and a European region.
Q2 was about structure — how apps, workspaces, and organizations fit together. Q3 is about what you can do inside that structure, and about the guarantees underneath it.
Four things stand out: a collaborative Assistant for designing and testing apps, a clear answer to who is accountable for every user, a rebuilt Insights, and Playlab that speaks and listens.
## What's new
A workspace that connects your apps, standards, and materials. Build an app from an idea, then role-play a student against it to see how it holds up.
Every user now has one organization accountable for them. Activity and safety flags follow the person, not the app they happened to open.
Redesigned and consolidated. Organization and app views in one place, scoped to the members your organization is responsible for.
Natural-sounding read-aloud in your chosen language, accent, and voice. Pick from a large catalog and set the playback speed.
## What's new for your role
### For builders
The Assistant changes how an app gets made. Describe an instructional idea, have it built, then stress-test it by simulating a student — a shy 7th grader who gives up quickly, say — and watch how your app actually responds before a class ever sees it. Agentic search now turns on automatically for any app with references, so retrieval works without a toggle.
* [Playlab Assistant](/features/Playlab%20Assistant)
* [Improving referencing with agentic search](/features/MCP%20RAG)
* [Python tool](/features/MCP%20Python%20Tool) and [document editor](/features/Doc%20Editor)
### For educators
Playlab now speaks and listens. Students can talk to an app instead of typing, and have responses read back in a natural voice in their own language and accent — useful for emerging readers, language learners, and anyone who finds a screen tiring.
* [Voice output](/features/Voice%20Output)
* [Voice input](/features/Voice%20Input)
* [Reviewing sessions](/getstarted/Reviewing%20Activity)
### For administrators
You can see what your people are doing wherever they work on Playlab, because activity follows the person rather than the app's owner. Insights was rebuilt around that scope. You can also restrict which AI models your builders may select.
* [Responsible organization](/features/Responsible%20Organization)
* [Organization insights](/features/Org-Level%20Insights)
* [Selecting an LLM](/features/Selecting%20an%20LLM)
### For network owners
Playlab now runs in a European region, with regional routing for models, embeddings, and voice, and an AI disclosure that meets the EU AI Act. Microsoft accounts work for single sign-on alongside Google.
* [Single sign-on](/faq/Single%20Sign%20On)
* [Privacy and security](/faq/Privacy%20and%20Security)
## A closer look
### An assistant that tests your work, not just writes it
Most AI helpers generate. The Playlab Assistant also checks. Give it a persona and a goal and it will hold a full conversation with one of your apps, playing the student, until the goal is met or the conversation runs out. Nothing touches your activity data — the simulation runs in memory — so you can run it repeatedly while tuning, and tell whether a change actually fixed the behavior.
It reaches official state standards in Math, ELA, and Science, open curricula including Illustrative Mathematics 360, OpenSciEd, Bookworms, and the OER Project, and released assessment questions from Texas, New York, Illinois, and Washington, D.C.
[Read about the Assistant →](/features/Playlab%20Assistant)
### One organization is accountable for each person
Previously, when someone used an app built elsewhere, it was ambiguous who should see that activity or receive a safety flag. Now every user has exactly one responsible organization — usually their school or district — and activity, flags, notifications, and exports all follow it. You don't set this up; Playlab assigns it.
[Read about responsible organizations →](/features/Responsible%20Organization)
### Insights, rebuilt
Organization and app insights were consolidated into one redesigned surface, scoped to the members your organization is responsible for. Builders see how their own apps are used; admins see across the organization.
[Read about Insights →](/features/Org-Level%20Insights)
### Playlab reads its answers aloud
Voice output reads responses aloud in a natural-sounding voice, with a large catalog of languages, accents, and voices, plus playback speed. Your browser's built-in voices are always available; the improved voice is an organization opt-in, because generating that audio means sending text to a third party. Voice input, which transcribes speech into the chat box, remains available separately.
[Read about voice output →](/features/Voice%20Output)
## Also shipped
* **Claude Sonnet 5 and Claude Opus 5** are selectable in any app's model picker.
* **Organization model control** lets admins restrict which models builders may choose, for state-law compliance.
* **Microsoft single sign-on**, alongside Google.
* **A European region**, with regional routing for models, embeddings, and voice.
* **Deeper tool use**, so apps can chain more tool calls and references without long generations failing.
* **Spaces** replaces "workspace" in the interface, and **Sessions** replaces "Activity."
## Coming soon
Moderation tuned specifically for the Assistant, rostering improvements, and continued expansion of the standards and curriculum library.
For everything as it ships, see the [product changelog](/changelog/Product%20Changelog). For the previous release, see the [Q2 2026 update](/changelog/Playlab%20Q2%202026%20Update).
## Learn with us
Live walkthroughs for new and returning users.
Drop in with questions any week. We're here.
## Contact us
Reach out any time at [support@playlab.ai](mailto:support@playlab.ai). For deeper conversations, you're always welcome at \[office hours]\(/learn with us/office hours).
***
Last updated: 09-22-2026
Contact us at [support@playlab.ai](mailto:support@playlab.ai)