Top Requested Features

[imtuyethan]

Here are the top requested features Jan has received in the last couple of months! Feel free to comment & add in folks. You tell us our roadmap.

Feature	Github link
RAG	https://github.com/janhq/jan/issues/3499 #3396
Settings preset / templates	#2845 #3508
Context token length	#2320
Vision	https://github.com/janhq/cortex/issues/208 #2575 #2828 #2824
Web-search	#3059
Threads Management	#3133
Knowledge Vaults	#2761
Agents	#2255 #2820	- User wants to switch between different assistants - User wants to leverage the assistant for creative writing, story development, character management, etc. - User wants customize and personalize the assistant's behavior, capabilities, and personality to better suit their individual needs and preferences
Voice (TTS & STT)	#3531
Hotkey & shortcut		- User wants to customize keyboard shortcuts
Edit responses	#2842	Use case: - Model can stick with the correct context / “Trim the fat” - To co-write - Edit style so the model can stick with
Multiple models		- See many models response at the same time for comparison - Want to run multiple models via the API
Fast model integration		I've noticing almost each posts on locallama, AI models are still in an experimental era - users are eager to try new models. Fast integrated with smooth experience wins on retention numbers.
Integration with popular tools in the market, like Ollama, VSCode,...	#2998 #2281 janhq/docs#91 #2831	Building a full-stack ecosystem is perfect/ideal for long-term run, but retaining users also depends on easy integration with popular tools in the market, like Ollama. Almost each post people're asking this.
Inline AI, quick ask options	#2757	Reaching AI as soon as possible. Features like Inline AI, quick ask options, and a small hand-waving logo on the right-hand side of the screen make Jan visible and readily accessible
Hardware compatibility (support AMD, better detection & utilization,...)	https://github.com/janhq/jan/issues/3500 #3529 #3530 https://github.com/janhq/cortex/issues/928 https://github.com/janhq/jan/issues/2630 #1525 https://github.com/janhq/cortex/issues/462 janhq/docs#15 janhq/cortex.tensorrt-llm#29 https://github.com/janhq/jan/issues/2813 https://github.com/janhq/cortex/issues/521 https://github.com/janhq/cortex/issues/522 https://github.com/janhq/cortex/issues/365 https://github.com/janhq/cortex/issues/500 https://github.com/janhq/cortex/issues/527 https://github.com/janhq/cortex/issues/805 #2587 #3360 https://github.com/janhq/cortex/issues/323 https://github.com/janhq/cortex/issues/677 https://github.com/janhq/cortex/issues/678 #3394 https://github.com/janhq/cortex/issues/1019	- Lack of support for certain hardware configurations (e.g., AMD CPUs, AMD GPUs with Vulkan) - Issues with model loading and performance, especially for larger models, on certain hardware setups - Need for better utilization of available hardware resources (CPU, GPU, RAM) to improve performance

[areyoufeelingitnowmrkrebs]

Bump for better AMD GPU support.

[ifr1m]

I noticed that ChatGpt (on MacOs at least) allows sourcing data from existing running programs: https://help.openai.com/en/articles/10119604-work-with-apps-on-macos#h_b2eb252f05

Do you guys think this would be feasible for Jan?

I am more than happy to help with this, if you think is doable :D

[AndroidHemroid]

Here are some suggestions from an old early 1980 programmer from back in the days for future features to consider implementing in the Jan interface:

1. Chat Folders: Introducing the ability to create folders for chats would significantly enhance organization. Users could label folders allowing the categorizing of chats into the folders, making it much easier to find and access specific messages and keep everything in order.

2. The Ability to save/backup all the chats along with the folders in the interface they were placed in by the user and the ability to import them back in along with the folders they reside in. This would allow users to have Jan on all their devices and be able to move back and fourth between them with all the same chats.

3. External Chat Storage Integration: Adding the option to specify a custom drive and folder for chat storage would enable users to save their conversations on external USB drives etc. This feature would allow for easy transfer between Jan installations on different systems and provide the capability to access chats directly from the external drive.

These enhancements would provide users with greater flexibility, organization, and accessibility, ensuring seamless chat management across the devices and platforms they use. Additionally, this could encourage more users to adopt Jan on multiple systems.

And if you really want to go above and beyond all the other Ai-IDE's add a feature in the settings that would allow all Chats at the beginning to have a time and date stamp those chats were made or edited last. And if wanted that could also be used to help build an incremental backup of the chats without having duplicates and allow for multi jan installs to only grab the new chats that are not on the other systems.

These features would significantly improve the user experience, making it more efficient, convenient, and enjoyable. Ultimately, they would represent a noteworthy advancement for the Jan application compared to other options like GPT-4All and WebOi running inside Docker.

[Culiex]

Math rendering would be greatly appreciated. Trying to read "[ x = \frac{-b \pm \sqrt{b^2 - 4ac}}{2a}" is just not comfortable.