Feature Requests

Have something you would like to see in Taskade? Let us know!
🔥 Granular AI Knowledge Control: Customize Modes (Only Knowledge, Non-Knowledge, Mixed) – Inspired by Google NotebookLM
Description: We propose adding a three-mode setting to Taskade’s AI functionality, allowing users to control how the AI interacts with their uploaded knowledge base (e.g., documents, notes, or project data). The modes would include: "Only Knowledge" Mode : Restrict AI responses exclusively to the user’s uploaded documents/data. The AI would not access external or general knowledge. "Non-Knowledge" Mode : Disable the AI’s access to uploaded data entirely (current "Disable Knowledge" setting). Responses would rely solely on the AI’s general training. "Mixed" Mode : Combine both the user’s knowledge base and the AI’s general knowledge (current "Enable Knowledge" behavior). This feature is inspired by platforms like Google NotebookLM , which prioritizes grounding AI outputs in user-provided sources. --- Why This is Important: Control & Precision : Users in regulated fields (e.g., legal, healthcare) need strict adherence to internal data. "Only Knowledge" mode ensures compliance and reduces hallucinations. Creativity vs. Accuracy : Writers/marketers might prefer "Non-Knowledge" mode for brainstorming, while teams needing data-backed answers can toggle "Only Knowledge." Flexibility : A hybrid "Mixed" mode remains useful for general tasks, but granular control empowers users to tailor AI behavior to specific workflows. --- How It Should Work: Accessible Settings : Add a dropdown menu in each AI Agent settings (e.g., under “AI Agent - Edit Agent - Knowledge”) to select between the three modes. Clear Labels : - 🔒 Only Knowledge : “Answers sourced 100% from your uploaded documents.” - 🎨 Non-Knowledge : “General AI creativity (no internal data used) (current default).” - 🌐 Mixed : “Combine your knowledge + AI insights.” Error Handling : In "Only Knowledge" mode, if a query lacks relevant data, the AI responds: “No matching information found in your workspace.” --- Use Cases: A legal team uses "Only Knowledge" to ensure all AI-generated contract clauses align strictly with their templates. A marketing team toggles to "Non-Knowledge" for brainstorming campaign slogans without data bias. A project manager uses "Mixed" mode to cross-reference task deadlines (from their calendar) with general productivity tips. --- Additional Notes: This aligns with Taskade’s vision of customizable workspaces and AI-augmented productivity. Competitors like NotebookLM emphasize source-grounded AI, making this a strategic differentiator. --- Closing: Thank you for considering this request! Implementing granular AI modes would significantly enhance Taskade’s adaptability for diverse teams and use cases! 🙌
7
·
in progress
Support Voice for AI
This feature request has a few features built into it. Support sending voice messages to AI and having it process and understand that information. So, when we're chatting with the AI agent, have the option for us to send a voice message instead of text. You can actually talk with your project data. Chat with the AI Agent through voice. So, you message it or talk with it and it'll respond back through voice. This is using this functionality from OpenAI: https://openai.com/index/chatgpt-can-now-see-hear-and-speak/ Transcription: Adding the ability to process voice would open the door to features like speech to text, uploading videos and audio files, and meeting summaries for meetings in Taskade. It unlocks a lot of new use cases and functionality. Eventually, given the other feature requests and functionality are supported, Taskade can handle user journeys like this: Let’s say you’re using Taskade for your company and small business. We’ll start with one use case: meetings. You ask it to find a time to meet with your team. It drafts the agenda, invite email, and sends it to you and your meeting attendees. While on the meeting, it transcribes and summarizes the meeting and users can ask questions live. After the meeting, it sends the meeting transcript and action items to every attendee. You can reference the meeting to send a follow-up email to the people that were on the call and you can have it suggest scheduling a follow-up on certain items. It creates the list of follow-up tasks for you based on the meeting and you can add it to whatever project you were working on for that meeting. If it processes any tasks like “set up a follow-up call” it will suggest doing that for you. You can then have it schedule it for you on your planner and calendar based on those tasks. When it schedules the follow-up, it’ll reference unfinished tasks from the previous meeting, your projects, etc. for drafting the agenda and invite email.
17
·
in progress
Load More