With the rise of AI macromodels, many people want to have their own AI Assistant, recently I found aAI Application PlatformThe ability to build your own AI apps quickly and with 0 code throughout.
The platform is calledbutton( https://www.1ai.net/2490.html), a platform for building AI applications.
I made an AI reading assistant because I usually love to readPublic AccountArticles, the longer articles are half an hour to read, it's very time consuming, so I thought of letting AI help me to extract the key points and help me to read quickly, so there is this AI reading assistant.
- Give links to articles that automatically summarize the core content for you to improve reading efficiency.
- AI can be asked to create different copy based on the content of the article twice, e.g., Little Red Book copywriting
- You can communicate directly with the article through dialog to increase reading depth.
realization
Go to the main page of Buckle and click on [Create Bot], and enter the AI app name
You will then see the AI app configuration page
- Character setting and response logic
- Skill
- commissioning area
First of all, enter your requirements under [Persona and Reply Logic], for example, enter "AI reading assistant, give you a link to a public number article, you help me summarize the content", and then click "Optimize".
It will automatically understand your needs and help you to improve, the more detailed your needs are described, the more accurate its understanding.
Next in the [debug area] we tested the results. I gave it an article about AI's big model participation in a math competition, and it ended up making up a nonsense answer for me!
This is because it doesn't have the ability to read links by default, you need to configure a plugin to read links in the [Skills] area.
We click "Add Plugin", you can see that it has a lot of built-in plug-ins, among which there is a "Link Reader" plug-in is what we need, click "Add".
Then we'll test it again with the same link and see the results!
Answering correctly, we can see that it differs from the previous process in that it first utilizes the plugin's ability to read the content of the article and then makes a summary based on the capabilities of the larger model.
We've got the effect we want for now, so we can just click 'Publish' in the top right corner.
Just need to authorize and configure it to distribute inside beanbags, flybooks, publics, nuggets.
The whole process 5 minutes to get it done, of course, I have not explained the function of the buckle comprehensive, I just design to a small part of the skills inside: workflow, knowledge base, database These advanced operations, interested partners can go to see the official document, more comprehensive.
The platform as a whole is very playable and very friendly for AI app entrepreneurs to quickly implement MVP ideas, making the cost of being wrong much lower.