Knowledge made accessible on Cardano.

Discord integrated AI to learn knowledge about your community and answer questions in real time.

Get Started

The AI Agent for your Discord community

It learns and improve the knowledge base over time, and you have full control over it!

Let community members ask questions in your Discord server and get answers from past questions.

Get Started

To get started, invite Mesh AI to your Discord server:

Invite Mesh AI

After you've invited Mesh AI to your server, you'll need to initialize the bot. To do this, run the following command in a channel that Mesh AI has access to in your Discord server:

!initmesh

Mesh AI will respond with a welcome message! You can now start using Mesh AI in your Discord server.

Ask any questions

To ask any question, just type the command:

/ask

Users might have to type / and then select the /ask command from the list of auto-suggested commands.

Mesh AI will respond with Embedded message containing the question. After a few moments, Mesh AI will respond with an answer to the question.

Fine-tune knowledge base

Mesh AI learns from your community and answers questions in real time.

There are 3 ways to fine-tune the knowledge base about your community.

1. Provide training data

You can provide training data to Mesh AI to help it learn about your community. To do this, run the following command in a channel that Mesh AI has access to in your Discord server:

/train

Mesh AI will respond with a message asking for the question and the answer. Providing both will create a question and answer pair in the knowledge base.

2. Reply to Mesh AI's message and react to it

First, you reply to Mesh AI's message with the correct answer.

Then, Mesh AI will add a 👍 reaction to your reply.

If you react with one of the following (positive) emojis on your reply:

👍 ✅ 💯 🔥 ❤️ 🙌 💪 🙏 👏 👌

Mesh AI will add your reply into the knowledge base.

3. React to a user's message that replied to another user's message

If you want to add a user's reply into the knowledge base, you can react to the user's reply with one of the following (positive) emojis:

👍 ✅ 💯 🔥 ❤️ 🙌 💪 🙏 👏 👌

Mesh AI then will add a 👍 reaction and add the reply into the knowledge base.

Note that, these are a list of negative emojis that you can use to indicate that a response is bad:

👎 🥶 😱 😵‍💫 😵 😡 🤬 🤮 🤢 👿

Implementation Roadmap

The current state of development of the Mesh AI is as follows

TitleDescription
✅ Integration Of LLM With Discord Part 1This would involve connecting the LLM to the Discord API, allowing it to be used on multiple Discord servers within the Cardano and SingularityNET ecosystem.
✅ Build Data Pipeline At BackendChat messages deem as potential training data are collected and organized
✅ Build ML Ops Pipeline Part 1From data to training data to model training.
✅ Training The ModelThe LLM would need to be trained on Cardano and SingularityNET-related topics using data from Discord chats. This would involve collecting and pre-processing the data, and then using it to train the model. Note that this is ongoing process and this work will last as long as the service is up.
✅ Build ML Ops Pipeline Part 2From model training to response serving
✅ Integration Of ChatGPT With Discord Part 2For serving responses and collecting data feedback.
✅ Testing And DebuggingThe integration would need to be thoroughly tested and any bugs or issues would need to be identified and fixed. Note that this is ongoing process and this work will last as long as the service is up.
✅ DeploymentOnce the integration has been tested and debugged, it can be deployed to the Discord servers where it will be used
✅ User Documentation And TrainingDetailed documentation and user guides would need to be created to help community members understand how to use the service, and training would need to be provided to help community members get the most out of the service.
Maintenance And SupportAfter the service is launched, it will need to be maintained and supported to ensure that it continues to function properly and to address any issues that arise.
Monitor And Measure PerformanceRegular monitoring of the service performance and usage is important to measure the success of the service and make any necessary adjustments.

Supported By