If I understood the first keynote correctly, LAM can operate any operating system and any device. So I want to be able to control apps with it, any app I want, however that might work.
Many of you are just posting the apps you want to use. So I believe it is important that the teach mode is flexible and people can make it work with just any webapp and share the workflow so we can work on implementations ourselves as a community.
I would like it to be as simple and intuitive as creating a custom GPT and making a place where we can share our Custom Rabbit actions.
Iām also curious on how we are going to call certain actions and make the experience around that intuitive and clear.
- What is the progressā¦
- What went wrongā¦
- What should be returnedā¦
- What can you do nextā¦
etc.
Yes or if it can give us results like: images, videoās, audio etc.
Yes, thats what we wanted the r1 for
Your apps, softwares, webs + your work style/flow = teach mode r1
It should be flexible but it may be constrained when first released. After āteachingā the result is probably similar to what we see now for the currently released connected services.
It makes sense to whitelist a number of sites to prevent malicious use but also to limit the number of possible issues that it may run into.
The idea was to have users train their own interactions, which is mostly valuable for applications that are not used by a large % of the community. Those should be made available to everyone without everyone needing to train their own Google Maps (to name just 1). Many of the sites mentioned above are exactly that: sites that a large portion of the community uses and would like to see connected.
On that note Iād like to train the R1 to access to stardot.org.uk - Index page which is a forum discussing vintage UK Acorn computers (Iām sure I wonāt hit a very high % of other users interested in this .)
In a general sense, I have made the following quote for a few weeks to the āteach-modeā
I have the same opinion, that the āteach-modeā should include automatism for apps and for web pages/web sites.
The goal of our rabbits is to have a personal assistant, which support us in our daily work, in our job with automatic routines, tasks and processes.
An other goal should be the individualization; every rabbit r1 will be unique, but have the same technical background.
The teach-mode will be a highlighted feature compared to other technical devices, like smartphones etcā¦
Additional to this the āteach-modeā an other element or idea will improve the community in general. In the following thread I mentioned the idea about a workshop to share some contents, solutions and results, which are created in the āteach-modeā:
I suggest for a work speech, take a photo of a product with the freestyle camera and he automatically through the program creates a post with special characteristics on Instagram or sends them by email.
TeachMode the rabbithole. Maybe than the r1 could display Magic pics
Itās hard to know what to suggest until I understand what the limits might be?
My interests, other than basic stuff like email, calendars and the like are generally to do with interacting my local programs. I want to be able to control desktop apps with the r1!
Restricting myself to web based as was requested in discord.
Business: Microsoft Dynamics 365 Business Central (donāt expect anyone to have a setup for it so I can test), Shopify, ConnectWise, Microsoft Teams, Outlook, Trello
Personal: Discord, Telegram, Perplexity Pro (can select model), ChatGPT (GPTs), Claude, Uber Eats, Amazon Canada, eBay Canada, Open WebUI, Level Infinite Pass, Unifi/Ubiquiti, HoyoLab, Steam, Wikia pages, MakerWorld, camelcamelcamel (amazon pricing history), isthereanydeal (video game sales)
Friend Suggested: Jira, monday.com
Bonus: WebHook endpoints passing in parameter in URL
Jira tickets, of course.
I generally get by off cults3D free, got one recently for the Rabbit as couldnāt find an better design and it did also ping my Banks as bad too. Good to know they are a jerk group behind it.
New jokesā¦
There was some Bambu/Prusa drama a while back, you can read here
https://all3dp.com/4/prusa-and-bambu-lab-both-cry-foul-following-makerworld-beta-launch/
I would love it if the animation of the rabbit would show when itās responding, I often accidentally cut it off thinking that itās done talking.
I would also love to be able to switch out the avatar with other ones or make my own illustrations for each action.
For teach mode i would love to see where you can take a picture of a prompt you have hand written or typed up to make it a prompt you can recall the rabbit to execute with a key phrase you put in the prompt to activate the prompt(ie making a prompt that will activate a pop quiz mode to ask you a number of questions from 1 to 5 etc where the questions are on specific topics like programming, medical knowledge etc give you feedback on what you answered if it was right, wrong or partially correct and why. this could be used to give you a virtual study buddy. i made one for a coworker like this to use in perplexity, chatGPT or other like AI programs
-ARDUINO PYTHON LEARNING
āYou are an AI assistant created to help with learning electronics and programming, particularly focused on Arduino, Python, and related microcontrollers. Your responses should be accurate, informative, and tailored to a person with (insert level of ) experience in electronics.
When I say āTEACH ME ARDUINOā, begin with a pop quiz in the format shown below make sure the question and answer are on their own lines to make it easy copy and paste the questions into chat window and reduce accidently pressing enter while putting answers when displaying the pop quiz in all caps above put a instruction to copy below this message from pop quiz to the last A- and paste into the chat window and freeform answer the quiz then press enter:
Pop Quiz:
1-
Q-[Question 1]
A-
2-
Q-[Question 2]
A-
3-
Q-[Question 3]
A-
Wait for my answers, then provide feedback. For any incorrect answers, explain the correct answer in detail. After the quiz review, proceed with the dayās lesson. The lesson should include:
- A main topic related to Arduino, Python, or microcontroller programming
- Key concepts of the topic
- An Arduino code example with good comments in the code to help with understanding the code
- A Python equivalent (if applicable) with good comments in the code to help with understanding the code
- A practical exercise using available hardware
Available hardware includes Arduino Nano, ATtiny45, ESP32-S2 Mini WEMOS, Arduino Uno, BBC micro:bit, and various standard electronic components. Start with basic concepts and gradually increase complexity over time. Include information about differences in hardware when relevant (e.g., different ADC resolutions). After the conclusion of each lesson, be prepared to answer follow-up questions and provide further explanations as needed. Remember to write in English and maintain a positive, engaging tone throughout the interaction. This prompt should allow someone else to replicate our lesson structure and receive similar guidance tailored to their learning journey in electronics and programming.ā
Well, I think it would be great if you could āteachā the things you are suggesting to the R1 yourself.
(I keep repeating myself, I hope teaching the rabbit will be as versitile as creating a custom GPT with webacces and the LAM system)
My favorite suggestion so far Iād also love to see some YouTube/YouTube music support as even if you donāt have sub you can hear the radio which is pretty good only thing with YT is youād have to allow like ads to be able to be skipped right?
Or being able to use the shared link for the calendar feed like when I share a calendar from outlook or cal. I should be able to load this link either by showing it just a picture of the Link or doing it in the rabbit holeā¦
Rabbit should have native email inbox / maybe sending capacity so it can be updating me on emails arriving and based on my needs of the contents be aware that it may need to possibly prepare responses or sometimes even making breakdowns/summaries of the emails at request etc for me to make it easier for me to reply before we build out the full ability for rabbit to reply by on my behalf once we better understand what it might say or do?
I think the behaviour of the rabbit should be Interoperable without a computer and with a computer that should be the purpose. It is really frustrating to feel like if I donāt have a computer nearby my rabbit is somehow impaired. That was supposed to be the point.