🤔 Why

Many people have similar views on AI to [JARVIS](https://en.wikipedia.org/wiki/J.A.R.V.I.S.), the personal assistant invented by Iron Man. They hope that AI can be a "personal super intelligent assistant" that can replace humans to do many things. However, I don't want future AI to be like JARVIS because it feels like a privilege for the rich and is both aloof and distant.

Instead, I hope future AI can be like Doraemon:

• It should be approachable, even for middle-class families like Nobita Nobi.
• In addition, it can be integrated into your life, work and study, existing like a family member, friend or teacher, rather than a tool with a high personality.
• Finally, it does not help humans by "replacing" them, as so-called intelligent assistants do, but by "providing" tools/methods to help them in any way possible.

In summary, "an assistant that enhances human intelligence" would match the ideal image of AI in my mind.

However, reality is harsh. Ever since I started using ChatGPT, I have found it to be a useful tool. However, during the course of my usage, I have come to realize that although it boasts of powerful capabilities, unlocking its full potential can be challenging and comes with its own set of difficulties.

For example, to improve the accuracy of the AI's answers, you need to learn a lot of skills. I spent more than 10 hours learning about Prompt Engineering to get the AI to return a satisfactory answer. But then I ran into a second problem, entering the Prompt information is very troublesome. It requires frequent copy-and-paste, which is very inefficient.

Lastly, payment problems may arise, just like with my experience. The monthly subscription fee is still too expensive for many people.

Since the product is not satisfactory, I tried to find a good solution on the market. Unfortunately, I tried various tools, and many of them were great, but none of them met all my needs. Therefore, we decided to develop our own.

😎 Differences

The development of AI is still in its early stages, and many technologies are not yet refined. There is still a long way to go to achieve our goals. However, this does not mean that we cannot take the first step towards making AI more approachable:

1. Lower the threshold of AI usage:
   1. Improve efficiency: Follow the product design that users are familiar with to improve the efficiency of using prompt.
   2. Lower cost: Support Pay as you go payment mode, pay as much as you use.
   3. Lower model invocation threshold: You can invoke multiple models at the same time in our application, such as GPT-3 and Stable Diffusion at the same time, and we plan to support Midjourney in the future.
2. Improve the accuracy of AI responses:
   1. Support calling external APIs: Because of the nature of the model, it cannot use the latest data to answer, such as "recommending the latest movie", etc. In the future, we will support calling external APIs so that the AI can use the latest data to answer.
   2. Custom answer data source: In some vertical fields, AI answers are often random, we will support custom database function, so that AI can answer based on the content in the knowledge base and improve the accuracy of AI answers.

🚀 Features

You can check the development progress of the following features through PoleStar Chat - Public Roadmap , and feel free to leave us feedback our Discord server.

IM App Design: To make the product more approachable, the product's design adopts IM interaction. You can create a chat room (Channel) and talk directly to the AI assistant through the input box like using other IM products. Also, to solve the problem of difficulty in calling prompt templates, you can quickly call prompts through the AT method, like this:

Support multiple models: Most products support only one model, but we support multiple AI models, such as text-to-image models or open-source models(Plan to support Midjourney and more!):