Sometimes when loading from scratch, it takes some time (2-3 curl requests) until it resolves the DNS of the Supabase instance

Some people will be more willing to pay for something than donate.
I would like to have an option where I paid more in exchange for doing less work.
That includes acquisition of hardware, installation of software, and maintenance of software.

If you didn't want to do this, would you be comfortable with someone else doing it?
Would you be willing to link to them in the readme?

much cheaper, and should do the work too

Post successful login, we aim to redirect users to the homepage using router.push("/"). This works flawlessly for users who have previously logged in. However, for new users, despite the login being successful (verified by the success toast message), the router.push("/") call does not redirect to the homepage.

We're considering if the issue could be bypassed by not relying on the router for redirecting post-login, especially since our homepage's rendering logic already accounts for user authentication state to display the correct view.

Otherwise maybe login should be its own page apart from the homepage

Currently the setup is without battery, need to find the easiest way to add this as part of the setup

Do I need to purchase an additional microphone module when deploying with Raspberry pi zero? I found that it doesn't seem to be stated in the documentation, and I'm confused about this

Relates to #1

We don't really need to compile everything separately - we can have official release here on github action

Summary:
Improve the UI to allow users to manage different AI models and configure corresponding API keys from the settings. Add a dropdown menu in the main chat interface for seamless switching between active models. Integrate persistent storage options to save these configurations.

Problem:
Currently, there is no integrated method for users to manage multiple AI models and their respective API keys within the application settings, nor a mechanism to persist these settings for future use. Users also need a convenient way to switch between active models directly from the chat interface.

Solution:

1. Develop modals in the settings section for model and API key management, allowing users to add, configure, or remove models and their API keys.
2. Implement a dropdown menu in the chat interface to switch between configured models easily.

Design:

• Modals within the settings for AI model and API key management.
• Dropdown menu in the chat interface for active model selection.
• Consistent and intuitive UI design that aligns with the current interface standards.

Persistent Storage Solutions:

• Integrate Capacitor for local storage of model configurations and API keys.
• Provide an option for cloud-based storage with Supabase for users who prefer to sync their settings across devices.
• Ensure that storage of sensitive information like API keys is secure and encrypted.

User Stories:

• As a user, I want to manage different AI models and API keys within the settings, and have these settings persist across sessions.
• As a user, I want to switch between configured models quickly from the chat interface for different tasks and comparisons.

Acceptance Criteria:

• Users can add, configure, and remove AI models and API keys within the settings modals.
• The dropdown menu allows for quick switching of active models in the chat.
• Model configurations and API keys are securely stored and persist across sessions and devices.

When I delete the last chat from the sidebar, the "loading" mockup stays and flashes. We could add a conditional. if chats are empty then display "Start Chat" or something along those lines

Add the option somewhere in the frontend to "turn-on" recording, that will send the data to the backend - as somewhat a replacement for the hardware device, so people can start using it without having the hardware yet

Currently it is using Wi-Fi, which makes it not-so mobile (you can technically connect it to your phone hot-spot).

An alternative approach would either be:

• Bluetooth pairing with the mobile device
• Sdd a 4G card that will allow it to be completely independent

metadata,
voice signatures

When the page loads, there is a brief flash of the LoginComponent before the user's authentication status is confirmed. This occurs because the initial state of the user in our authentication hook is set to null, and it takes a moment for the hook to update this state based on the user's session. This flash can be jarring and confusing for users who are already logged in.

Steps to Reproduce

Open the application while already logged in.
Observe that the LoginComponent briefly appears before being replaced by the main content.

Expected Behavior

The LoginComponent should not appear if the user is already authenticated and has an active session.

Actual Behavior

The LoginComponent flashes briefly on page load, even when the user is authenticated.

Designing an identity for the first open source AI wearable device. This identity should include the logo, an identity system, app icon, epehemerals, social media post templates.

Hey I think a mobile app or desktop app might be just as useful.

We always have our phones with us and a lot of us spend time with airpods in.

Implement speaker diarization and VAD. This will let the agent understand who is speaking, providing the user with better responses. This should also get rid of audio hallucinations when there is silence, very important as the wearable will be recording during silence often if worn all the time.

Potential implementations:
https://github.com/pyannote/pyannote-audio
Deepgram

That one could be interesting for the device. It's cheap, has an inbuilt camera and microphone.

https://www.seeedstudio.com/XIAO-ESP32S3-Sense-p-5639.html

Add a button on the interface to stop the assistants response.

Render Markdown and LaTeX in the chat interface to improve usability, support lists, headers, equations, etc.

Seems like we will have growing menu items, and stuff like that.

Probably need to implement some kind of menu Items minimizable.

Found some cool examples

https://codesandbox.io/p/sandbox/framer-motion-side-menu-mx2rw

Can I set up this project using RespberryPi 4B?

The "setup hardware" link is a 404

https://docs.adeus.ai/setup_coralai

I want to flash my seeds studio c3 (as in the docs) but i get errors, regarding esp_ble_gap_ext_adv_report_t, e.g:

In file included from /Users/solar/Repos/private/ADeus/devices/esp32-c3/components/arduino/libraries/BLE/src/BLEClient.h:25, from /Users/solar/Repos/private/ADeus/devices/esp32-c3/components/arduino/libraries/BLE/src/BLERemoteService.h:18, from /Users/solar/Repos/private/ADeus/devices/esp32-c3/components/arduino/libraries/BLE/src/BLERemoteService.cpp:14: /Users/solar/Repos/private/ADeus/devices/esp32-c3/components/arduino/libraries/BLE/src/BLEAdvertisedDevice.h:148:31: error: 'esp_ble_gap_ext_adv_report_t' has not been declared 148 | virtual void onResult(esp_ble_gap_ext_adv_report_t report) = 0; | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~

Any ideas why?

I installed esp in 5.1.3 version and additionally i used the idf-release/v5.1 branch for https://github.com/espressif/arduino-esp32.git but with no success

My system is macbook pro m2, python 3.11 and i followed all the steps as defined in the readme and website

The current system prompt specifically references "He" when describing interactions with the assistant. To ensure our language is neutral and universally applicable, I propose we update the prompt to use "User" as a more general term.

Suggested Revision:

Change the existing prompt from:
"You are a helpful assistant, helping the user navigate through life. He is asking you questions, and you answer them with the best of your ability."

To:
"You are a helpful assistant, helping the user navigate through life. The user is asking you questions, and you answer them with the best of your ability."

supabase/functions/chat/index.ts

I got '400 The audio file could not be decoded or its format is not supported.' from logs.'
I test the program:
1.python_recorder_client on my local macbook record a good audio can playback by quicktime and post to supabase function got a success response from OpenAI API
2.The audio which recoded by coralai board ' s format is invalid. Can not play by quicktime and got '400 The audio file could not be decoded or its format is not supported.' from OpenAI API

This is directed more at the owner of the repository;

Do you plan on maintaining and and supporting this repository/project even with your implications in a direct alternative/competitor

Thank you;

Whisper tends to generate YouTube-like text when the audio is unclear, so you can get noise data in the database like "Thank you for watching", and "See you in the next video" - evem though it has nothing to do with the audio

Need to spend more time on the Supabase RLS, etc.

Currently, it is somewhat widely open