hope it will be available for self-hosted version 🙏🙏🙏
Tony Dinh changed status to In Progress
Team•
Aug 12, 2024
Tony Dinh
Team•
Jun 2, 2024
It's strange that you need to have a limit for each user.
But anyway, here is how you can limit the number of message characters per user, for example: user A can only have 50,000 characters per month. This can be done with the current user tags system.
Create a custom model.
Assign a limit to the custom model (e.g., 50,000 characters per month)
Create a unique tags for each user you want to assign the limit to.
Create an AI agent and assign the custom model to the AI agent, then limit the visibility of this AI agent based on the user tags.
Does this work?
PS: step 1 can be removed if you work out the rules in the model limit setting. You can make a setting to apply to only a specific user tags, but it can get complicated so I suggest creating a custom model for each user to simplify it.
Sapphire Horizon
•
Jun 2, 2024
Tony Dinh For my students I can indeed work with character limits Tony. Thank you for illustrating the steps. Stef
Sapphire Horizon
•
Jun 5, 2024
Tony Dinh After a lenghty discussion with VA today, herre is an explanation why Resellers likely will require token limits per user:
====
From a business management perspective, we need to have precise control over the cost of AI-consumption as we are bundling AI agents with our courses. Example: When we will sell an Intro-level course for lets say 147$, we include the access to AI agents in the pricing.. after the course ends, students switch to memberships at 37$/month (prices not final, just to give an idea).
Memberships provide continuous access to AI Agents after the course has ended. For monthly memberships, we need to make sure students dont go over a cerain $$$ threshold when using our TM AI agents (if a student pays 37$/monthfor access to AI, but uses 50$ of tokens, we are loosing money) - TM currently does not offer the possibility to limit users by token consumption - only character limits are currently possible. Its a workaround, but less precise.
*Example of pricing structure in a Reseller scenario:
Our Net profit before taxes: approx. 80$ (92$ - 12$ variable cost)
Net profit after taxes (Canada): 80$ - 35% = 52$
This example shows you the importance of limiting AI token consumption as precisely as possible - overconsumption would directly impact our bottom line.
WHY limiting tokens "PER USER": some users will require more, some less tokens per month when ona monthly membership plan. This means we would like to offer different AI-access packages for light- vs. heavy users with differnet pricing and differnt token limits.
Just today TheBelleStory posted a new feature for limiting agents by total token consumption. Thats GREAT! This would probably be the right place to add a feature extension that would allow to limit tokens based on a time-constraint: weekly, monthly, from-to [date] and total token consumtion, for example. TheBelle asked me to add this as a seperate feature request. Will do.
Cheers! I simply love what you and your team are doing, Tony! :-)
Stef (aka The Spiderman)
Yellow Flamingo
•
Jun 9, 2024
Stephan Becker good idea
Gold Orbit
•
May 17, 2024
Yes being able to adjust token limit per model per user per character is very important!’
Log in to comment and vote
Comments8
Nov 26, 2024
https://docs.typingmind.com/typingmind-custom/user-management/usage-limits:-set-limits-per-ai-model-agent-and-user
Turquoise Lake
Sep 5, 2024
hope it will be available for self-hosted version 🙏🙏🙏
Aug 12, 2024
Tony Dinh
Jun 2, 2024
It's strange that you need to have a limit for each user.
But anyway, here is how you can limit the number of message characters per user, for example: user A can only have 50,000 characters per month. This can be done with the current user tags system.
Create a custom model.
Assign a limit to the custom model (e.g., 50,000 characters per month)
Create a unique tags for each user you want to assign the limit to.
Create an AI agent and assign the custom model to the AI agent, then limit the visibility of this AI agent based on the user tags.
Does this work?
PS: step 1 can be removed if you work out the rules in the model limit setting. You can make a setting to apply to only a specific user tags, but it can get complicated so I suggest creating a custom model for each user to simplify it.
Sapphire Horizon
Jun 2, 2024
Tony Dinh For my students I can indeed work with character limits Tony. Thank you for illustrating the steps. Stef
Sapphire Horizon
Jun 5, 2024
Tony Dinh After a lenghty discussion with VA today, herre is an explanation why Resellers likely will require token limits per user:
====
From a business management perspective, we need to have precise control over the cost of AI-consumption as we are bundling AI agents with our courses. Example: When we will sell an Intro-level course for lets say 147$, we include the access to AI agents in the pricing.. after the course ends, students switch to memberships at 37$/month (prices not final, just to give an idea).
*Example of pricing structure in a Reseller scenario:
Course price: 147$
Whats included:
Our course management+seat management cost: 27$/month/user
TM seat: 8$/month
Token consumption: 20$/month max.
Our Gross profit: 147$ - 27$ - 8$ - 20$ = 92$
Our Net profit before taxes: approx. 80$ (92$ - 12$ variable cost)
Net profit after taxes (Canada): 80$ - 35% = 52$
This example shows you the importance of limiting AI token consumption as precisely as possible - overconsumption would directly impact our bottom line.
WHY limiting tokens "PER USER": some users will require more, some less tokens per month when ona monthly membership plan. This means we would like to offer different AI-access packages for light- vs. heavy users with differnet pricing and differnt token limits.
Just today TheBelleStory posted a new feature for limiting agents by total token consumption. Thats GREAT! This would probably be the right place to add a feature extension that would allow to limit tokens based on a time-constraint: weekly, monthly, from-to [date] and total token consumtion, for example. TheBelle asked me to add this as a seperate feature request. Will do.
Cheers! I simply love what you and your team are doing, Tony! :-)
Stef (aka The Spiderman)
Yellow Flamingo
Jun 9, 2024
Stephan Becker good idea
Gold Orbit
May 17, 2024
Yes being able to adjust token limit per model per user per character is very important!’