AI music, voice & developer tools · Product news & updates

AI news: Suno MCP beta, Google Connected Apps and Copilot harness — 29 October 2026

Suno opens MCP beta access; Google previews Connected Apps and adds immersive learning. Plus Copilot harness, Dot texting, ElevenLabs SDK and Grok API updates.

Publisher disclosure: MUVO offers competing music, remix, and video tools. This report summarizes public sources and provider documentation; it is not an independent product test or a claim that MUVO is free of similar issues. Read our reporting method.

Key product updates at a glance

Suno has announced an MCP beta for making music inside AI assistants; Google has opened a Connected Apps platform preview; Microsoft has introduced its Copilot harness in Foundry. The practical theme is connecting existing tools to agents, with important differences in access, costs and rollout stage.

This edition covers 26 October at 12:03 through 29 October at 12:08 Europe/Istanbul (UTC+3). Most announcements below are dated 28 October; individual dates are specified. Previously covered launches are not repeated as new releases.

Product What changed Access at cutoff
Suno MCP Music creation through compatible assistants Early-access beta; signup; all Suno plans eligible
Google Connected Apps Shared integration path for Gemini and Search Preview applications; phased access
Microsoft Copilot harness Managed agent execution in Foundry Public preview
ChatGPT Dot texting iMessage and Android RCS conversations Beta rolling out to US Pro users
ElevenLabs Python SDK Conversation disconnection details Version 2.75.0 released

Suno MCP beta: create music from an assistant conversation

Suno’s 28 October release notes announce an early-access MCP beta supporting Claude, ChatGPT, Claude Code and other compatible clients. It can create songs from descriptions or uploaded audio, and work with library songs through covers, extensions and mashups. Outputs return to the Suno library, with playlist actions available from the conversation. Suno release notes.

All Suno plans, including Free, are eligible, but this remains an early-access signup rather than automatic access for every account. Music creation consumes the account’s normal credits. Suno’s FAQ describes signing in and authorizing its official connector; a host assistant may require a custom-connector setup. Eligibility for Suno does not establish that every host plan supports that setup. FAQ updated 28 October.

For musicians, the benefit is keeping planning and creation in one conversation. It is not a new music model or a promise of unlimited generation. Check the host’s connector support and the Suno credit charge before starting a paid workflow.

Google launched the preview of its Connected Apps Platform on 28 October. Developers can expose tools through a cloud-hosted MCP server or Android AppFunctions, then register, test and submit the integration in Google Connected Console. They select Search, Gemini or both before review, and can monitor traffic, latency and errors after publication. Official announcement.

Access is phased and developers must apply for the preview. Google says partner integrations including Adobe and Peloton are live in Gemini, with expansion to Search planned over the coming weeks; Calm is also planned. This does not mean each integration is already available in every product, account or region. Developer overview and preview application.

The useful change for app developers is a common route for authorized account data and actions. A listed integration still needs its own review, authentication and user eligibility checks.

Microsoft Copilot harness: public preview, with a separate resilience release

Microsoft introduced the Copilot harness in Foundry Agent Service in public preview on 28 October. Microsoft manages the agent loop and sandbox; developers supply instructions, skills, tools and a supported model. The preview includes reusable skills, connected tools, files and code execution, bounded delegation, and source-controlled agent definitions. Foundry announcement.

The harness must currently be selected when creating an agent; switching it afterward is not supported. Microsoft also announces general availability this week of long-running resilience for hosted agents. That is a separate capability: interrupted work restarts with its original input, while application code must save progress and prevent duplicate external actions.

For developers, managed recovery can reduce hosting work, but it does not automatically make every workflow safe to replay. The preview and the generally available resilience feature should not be treated as having identical maturity or behavior.

OpenAI Dot texting: US Pro beta reaches iMessage and RCS

OpenAI’s 28 October update introduces one-to-one texting with a Dot through iMessage or Android RCS, rolling out in beta to ChatGPT Pro users in the United States. Users can send text, photos and voice messages; approval requests link back to ChatGPT. The documentation groups this update in the week of 26–30 October. Dated update, OpenAI staff announcement.

Setup starts from the Dot profile’s Texting option and requires phone-number verification. The contact method shares relevant conversation context and connected plugins. Group chats, calls and FaceTime through texting are not supported; setting it up does not grant access to unrelated text conversations. Current setup and limits.

This is a limited consumer rollout, not a general SMS API launch or confirmation of worldwide availability.

Google Gemini: immersive learning scenes are available, narration is still planned

Google added a library of interactive visual explorations to the Gemini student hub on 28 October. Users can move through layers of a topic, from large structures to smaller details. Google says all users can access the feature in English, Indonesian, Hindi, Brazilian Portuguese, French and simplified Chinese; other languages receive an English experience for now. Original announcement.

Access directly within a Gemini chat and audio narration are described as coming soon. Educators and creators can explore the current library, but should not assume those forthcoming features are already enabled. This is an interactive learning feature, not a separate Veo, Flow or Lyria model release.

ElevenLabs Python SDK 2.75.0 adds disconnection details

ElevenLabs’ Python SDK release adds conversation-disconnection reporting and regenerated API bindings. The GitHub release timestamp is 27 October at 23:02 UTC, which is 28 October at 02:02 Istanbul; the package files were uploaded to PyPI on 28 October. Release notes, published package.

For voice-agent developers, more explicit disconnection information can help diagnose why a session ended. This is an SDK change, not evidence of a new voice/music model, an incident recovery or a service-wide availability change. Review the linked code changes against the version your application uses before upgrading.

Grok text APIs gain content-guideline controls

The xAI API documentation, now branded SpaceXSI, dates custom moderation controls to 27 October. Chat Completions and Responses requests can include a guidelines array controlling how Grok handles the nsfw, violence and profanity categories as it writes a response. Dated release note.

This gives developers an additional request-level configuration point. The announcement concerns text APIs; it does not establish equivalent controls for the consumer app, video or voice products, or guarantee that every unwanted output will be blocked.

AI research access: Anthropic and Meta expand institutional programs

Anthropic’s 28 October announcement expands AI for Science through 20 foundations, institutions and research groups. The program distributes Claude credits for scientific work, including via partner organizations. Anthropic also highlights open datasets, software and standards as possible outputs; the Model Hardware Standard remains a research preview. Anthropic announcement.

Meta separately announced $2 million in grants and model access for five US universities on 28 October. The named partners are Caltech, Ohio State, Louisiana State, Indiana University and UTHealth Houston. Projects include computer vision and adaptive conversational systems for research in healthcare and science. Meta announcement.

Analysis: these are concrete changes to institutional access and research support, not a general free-account offer or proof of clinical benefit. The projects and vendor descriptions should be evaluated as research, with results and availability distinct from consumer product launches.

Sources

Original announcements, release notes and current eligibility documents are also linked beside each story. The source dates distinguish announcements, package publication and later documentation checks. MUVO has not benchmarked these releases or used paid generations to test them.

Previously covered Runway Ads, App Studio, Genie 4, Apple’s November creative updates and other overlapping announcements remain in the 28 October roundup and 27 October roundup. Sparse or inaccessible specialist feeds limit completeness; an absence here is not proof that a product had no changes.

MUVO publishes this reporting and offers competing creative tools. See our reporting method and the separate incident report.

Update timeline