Release 64 adds recurring tasks

Capacities is a note-taking app, but this update is really about automation. Release 64 adds recurring tasks and lets those automations use the AI model provider you choose, not just one default vendor. That sounds small. It isn’t. Basically, it pushes Capacities from “ask the AI something right now” toward “set up work that keeps happening in the background.” (youtube.com) ### What actually shipped? The(youtube.com)recurring tasks, and the ability to choose your Capacities AI model provider for those workflows. The first half of the video is dedicated to recurring tasks, and the second half to model-provider choice. That framing matters because it shows what the team thinks the product is becoming — not just a place to store notes, but a place to run repeatable systems. (youtube.co([youtube.com)e recurring tasks a bigger deal than they sound? A normal task feature helps you remember something once. A recurring task feature helps you run a routine. Those are different products. One is a checklist. The other starts to look like an operating system for your week — review every Friday, prep every Monday, publish every month, clean up every evening. Capacities had already moved into task management in Release 57 and added deadlines in Release 63(youtube.com)capacities.io) ### Why does model-provider choice matter here? Because automations lock in faster than chat does. If you build a recurring workflow around one model, changing providers later can be annoying, expensive, or both. Capacities started addressing that in Release 56, when it let users plug in their own API keys from OpenAI, Anthropic, Google, Mistral, or xAI, set spending limits, and choose where data gets sent. Release 64 appears to carry that flexibility into the automati(capacities.io) no longer welded together. (capacities.io) ### Why is that useful in practice? Different models are good at different things. One might be better at writing, another at reasoning, another at cost, another at privacy posture. If your recurring task is summarizing meeting notes, you may want one provider. If it is classifying research or drafting structured outputs, you may want another. The catch is that this only helps if the app exposes the choice cleanly inside the workflow itself — and Release 64 (capacities.io)ghted. (capacities.io) ### Is this just an AI feature? Not really. It is a product-architecture feature. Capacities had already been making its AI assistant more grounded and more agent-like in Release 59, with richer context, linked-note retrieval, and a roadmap toward an assistant that can research before answering. Recurring tasks fit that direction. They give the AI something durable to do on a schedule instead of waiting for a prompt every time. (capacities.io)oader market? Because AI software is moving from answers to actions. You can see the same pattern elsewhere — tools are racing to add background agents, task runners, and automation layers instead of stopping at chat. Capacities is applying that pattern to personal knowledge work. In plain English, your notes are becoming a trigger surface for routines, not just a place where information sits still. (microsoft.com)sks-from-answers-to-actions/)) ### So what changed for users? Before, the main AI interaction was episodic — open a note, ask a question, get a response. Now the shape shifts. Users can set repeating work and pair it with provider flexibility. That makes the app stickier, but it also makes setup decisions more important — cadence, cost controls, and provider choice start affecting everyday operations instead of occasional experiments. (youtube.com) ### Bottom line? Release 64 is less about flashy AI and more about dependable AI. That is the real shift. Capacities is turning repeated personal workflows into scheduled systems — and making the model underneath those systems something you can swap instead of inherit. (youtube.com)