AI is quietly moving out of the chatbot window and into the software people actually use to make things. In the past week, Apple added AI assistants to its professional creative suite and ElevenLabs shipped a new model that turns a single sentence into a finished sound effect — two very different companies, pointed at the same idea: let AI do the first draft of the work, not the whole job.
Apple Puts an AI Assistant Inside Keynote, Pages and Logic Pro
Apple announced on October 27 that its Creator Studio Pro subscription — the $12.99-a-month, $129-a-year bundle it launched back in January — is getting its biggest AI upgrade since launch. Keynote, Pages and Numbers now include a built-in assistant, powered by a mix of Apple’s own models and OpenAI technology, that can read a document’s existing content and draft slides, rewrite text or restructure a presentation around it. The feature is labeled beta and, for now, works in U.S. English only.
Logic Pro gets its own version on the Mac: a beta assistant that can sketch a song from a text description, build out vocal tracks from written instructions, or edit an existing project on request. It requires macOS 27. Apple says the update rolls out as a free addition for existing subscribers starting in November — no new tier, no extra charge — and that content generated inside Creator Studio Pro is not used to train its models.
It’s a notably narrow rollout for a company that tends to make AI announcements in sweeping keynote fashion. One assistant, one language, two apps-worth of beta labels — the kind of staged release that suggests Apple is still more cautious about shipping generative features into paid professional tools than its chatbot-first competitors are.
ElevenLabs’ SFX v3: One Prompt, A Whole Scene of Sound
Two days later, on October 29, voice-AI company ElevenLabs released SFX v3, the newest version of its text-to-sound-effect model. The pitch is specificity: where earlier versions of the model handled a single instruction reasonably well, ElevenLabs says v3 can follow several layered instructions in one prompt — a footstep on gravel, a door creaking open, and distant traffic, all described in one sentence — and actually produce all of them rather than picking just one. The company also says it improved prompts that blend music and vocals together, and cleaned up artifacts on metallic and resonant sounds like gongs, chimes and whistles that earlier models tended to smear.
SFX v3 is live across ElevenLabs’ ElevenCreative surfaces — the Flows tool, the dedicated Sound Effects tab, Studio and the Music Editor — as well as in the mobile app, and developers can call it directly through ElevenAPI using the model ID eleven_text_to_sound_v3. Generations top out at 30 seconds each, output as MP3 by default, with an optional 48kHz WAV for effects that don’t need to loop. ElevenLabs hasn’t published a v3-specific price; billing follows the existing sound-effects endpoint, which charges in credits per second of generated audio.
It’s a small-sounding update — sound design is a niche compared to chatbots or image generators — but it’s exactly the kind of incremental, workflow-specific release that has defined ElevenLabs’ approach since it built its name on AI voice cloning: pick one narrow production task, and make the model genuinely better at just that task.
The Pattern Underneath Both Releases
Neither of these is a flagship model launch with a benchmark chart attached. Both are AI arriving inside tools that creative professionals already pay for and already know how to use — a slide deck, a Logic Pro session, a sound-design pass in post-production — rather than asking anyone to learn a new chatbot interface first. That’s a different kind of AI news than the monthly parade of frontier-model releases, but it may matter more to how most people actually encounter the technology day to day: not as a conversation, but as a shortcut quietly built into software they already open every morning.
Apple’s rollout lands in the first half of November; ElevenLabs’ is already live. Expect both to get quieter follow-up updates — wider language support for Apple’s assistant, pricing details for ElevenLabs’ model — rather than another headline moment. That’s increasingly how AI ships in late 2026: not with a keynote, but with a changelog.
