AI Roadmap ​

We envision AI to be used in every layer of Capacities to make your workflows simpler, faster, and more powerful.

AI is not a feature, it's a fundamental part of Capacities. Integrating AI into Capacities is a project that will take many months to complete. We have a clear path for the upcoming months and a more general vision and fuzzy ideas for the more distant future.

Our vision with AI ​

We see enormous potential in facilitating AI to help us in our workflows. We envision AI to be a helpful assistant integrated into every layer of Capacities to make your workflows simpler, faster, and more powerful.

We also think that AI should always be a companion that helps us. Capacities will never become a "note-taking app on autopilot" but a tool that helps us to be more productive and creative. This means that AI should be used when it's helpful and should not when it's not.

Capacities AI today ​

You can get an overview of all AI features in Capacities here.

Roadmap ​

Here's what we're working on and what we have planned for the upcoming months. This is not set in stone and will change over time.

Near-term projects ​

• 🔳 Extended settings on how you want to use AI: Already now, you can decide to use AI or not. With the upcoming features, we need to process all your data with AI models. Our data will never be used to train AI models nor will it be stored on third-party servers. With the new settings, you'll be able to turn on and off the Capacities AI for your spaces separately. We'll also focus on making it more transparent what data is used and how it's processed.

• 🔳 Unlinked mentions: Sometimes we write about content we already have as objects in Capacities. We'll display unlinked mentions of an object below your backlink section. You'll be able to link them with a click.

• 🔳 Related content: Linking helps us to browse our network of thoughts. To help you in that process, we'll show semantically related content to the current object below your backlink section. You'll be able to start writing and Capacities will suggest related content. Capacities will become a true assistant that supplies you with relevant content.

• 🔳 AI editor: We'll rework the interface of the AI assistant in Capacities. Without leaving the editor you'll be able to quickly define a word, summarize a note, extract TODOs and much more. Capacities will provide commands you can quickly use in your workflows.

Long-term projects & more fuzzy ideas ​

• 🔳 Tag page enhancements: We'll suggest semantically related content at the bottom of your tag pages so it's not only the content you already tagged but everything that might be related. This will enrich exploration within Capacities.

• 🔳 Full-text search in media: We'll add full-text search to PDFs, images, videos, and audio files. This will allow you to search for text in media files and will make discovery within Capacities easier. We'll also rework extended search to include media search as well.

• 🔳 Unified search experience combining various methods of search: We want to integrate media, semantic search, and full-text search into one unified search experience so you can just search for anything and get the best results.

• 🔳 Understanding media and chatting with media: We'll add media content like text in images, audio and video transcripts, etc. to the context of the AI assistant. This will allow you to ask questions about the media and to use it in your workflows.

• 🔳 Chat with your whole space: We have a lot of super exciting concepts and ideas around this topic. We currently plan to extend the context of the AI assistant to tag pages, objects, collections, etc. and will then over time make the AI assistant capable of answering questions with the knowledge of all your content in a space.

Open questions and ideas ​

• ❓ Custom commands on property or object level: We're playing with the idea of allowing you to define custom commands for objects and the editor. While we always plan to incorporate many useful commands into Capacities, you might have custom workflows that require custom commands. We're also thinking about introducing a command library so you can easily share and use commands created by others.