TypingMind for Teams

Extract text from PDFs while inserting markers for actual PDF viewer page numbers (e.g., [Page 5] at the start of each page). Pass these markers into the prompt text, so the AI can accurately reference the same page numbers seen in a PDF viewer.

For the purpose of Digital Security, Legal, and internal KPIs, I would like the TypingMind API to be updated to allow admins access to pull chat conversations from TM storage databases.

Support voice custom model with different endpoint like whisper-1, tts-1, tts-1-hd. and possibilities to set up it instead of Web API and hard coded Openai Whisper

I'd like to request a feature that allows AI agents to trigger and communicate with other AI agents directly from within their system prompts. Currently, the platform allows users to tag an agent in the chat interface, but there's no native capability for agents to autonomously initiate communication with other specialized agents when needed. This would enable creating hierarchical or collaborative agent networks where: A primary agent could recognize when it needs specialized knowledge The primary agent could trigger a secondary agent with specific expertise The secondary agent would provide its specialized input The primary agent would incorporate this input into its final response Similar to how users can use "@" in the chat interface, agents should be able to reference other agents within their system prompts with directives like: If you need code information, request data from @CodeAgent This would extend the current multi-agent workflows to be more autonomous and embedded within the agent's decision-making process.

Hi, we are using Typingmind custom at our university. We would like the students to be able to change models, but we do not want them to be able to adjust model settings. Can you please set up an option where these two are separated? Thanks, Burak

Objective : Enhance TypingMind by adding long-term memory capabilities to AI agents, allowing them to retain and recall information over extended interactions, thereby improving user experience and personalization. Solution Overview : Integrate long-term memory using one of the following options: mem0.ai : A specialized memory platform for AI agents that offers seamless memory storage and recall, optimized for conversational AI. Integrating mem0.ai would allow TypingMind agents to retain knowledge across sessions without complex infrastructure. LangChain Memory : If TypingMind's tech stack is more aligned with Python-based frameworks, LangChain provides a variety of memory modules such as ConversationBufferMemory and EntityMemory , which can be customized to store and retrieve information dynamically. Implementation : Data Storage : Use vector embeddings to store conversation context or important information. Retrieval : Retrieve relevant memory chunks based on context to ensure seamless interactions. Tech Stack Compatibility : Evaluate whether mem0.ai or LangChain aligns better with TypingMind’s existing infrastructure (cloud, database, language). Recommendation : Depending on TypingMind's existing stack, either implement a pre-built memory system (e.g., mem0.ai ) or use LangChain's modular memory capabilities. Alternatively, the development team can research and propose a custom solution that fits their stack.

I find that every time I modify the default model on the admin panel, the change doesn't propagate to my users. I still want to allow them to switch to other models, so I don't want the "enforce default model" setting, but I DO want to be able to control what model is used when they press "new chat". Right now, I have to send a company-wide email blast everytime a better model is available and let folks know that they need to switch their default on their models tab. This is not ideal.

Integration with Confluence (Atlassian) to use as a knowledge base source.