Create a Flowise Assistant with Custom Tools and APIs | AI Automation Tutorial

Create a Flowise Assistant with Custom Tools and APIs | AI Automation Tutorial | Part 1

Рет қаралды 7,214

7 ай бұрын

#flowise #langchain #openai
Lets learn how to create a flowise assistant api using custom tools for a solar energy company. In this tutorial we'll create an AI automated assistant using Flowise to connect to Google sheets and also enable website visitors to calculate an estimated cost for a solar panel installation.
After seeing OpenAI's assistant where they could do function calling, we wanted to create a chatbot that could do the same thing! Instead of just answering qestiions, we wanted to create an AI assistant could independently call functions using the custom tool feature found in flowise.
It turns out that if you use an automation platform like N8N, you can connect to external APIs and data in a way that gives your AI assistant superpowers. Look up information on the internet, perform calculations and interact with website visitors in new ways.
Get Access to Chain Node Mastery (Flowise Mini-Course):
onyxstudiosinteractive.com/ch...
Need AI Consulting?
onyxstudiosinteractive.com/co...

Пікірлер: 25

@MarxOrx 6 ай бұрын

Bro you’re saving lives here 🎉

@ryanvaughnwilliams6113 6 ай бұрын

Thank you so much!! God bless you!

@GilbertMizrahi 7 ай бұрын

For the calc estimate tool, you could add the JS in the tool itself without having to send the data to a webhook where the calculation is done.

@OnyxStudiosInteractive 7 ай бұрын

Thanks for commenting. You're right, you could totally add the javascript directly in the tool itself. Although it is more fun to have the option of sending the data to a webhook and potentially making it accessible to other agents or adding some additional automation to it.

@user-zf1ur9lh7o 2 ай бұрын

Thanks for the great work. I really enjoyed the tutorial. Sorry or my question but I am new to integrating Flowise with N8N. I have an ec2 instance on aws that has n8n installed and I was trying n8n webhook with Flowise. Re the webhook method, if I use "POST", I get this error message ""This webhook is not registered for POST requests. Did you mean to make a GET request?""...... even I tried to use "GET".....but nothing works and it doesn't receive anything from the Flowise chatbot...... Can you help please?

@gautammandewalker8935 3 ай бұрын

Great tutorial! I am just curious to know how can we do the data collection in steps, like a questionnaire. It may get difficult for users to understand that they have to type name, email, kilowatt requirements together.

@OnyxStudiosInteractive Ай бұрын

You can always use a form to get the information and then pass the data to the flowise API. Check out the product recommendation engine video

@RolandoLopezNieto Ай бұрын

I was wondering how to setup an agent that supported a vector database + memory + custom tools. Thank you very much

@OnyxStudiosInteractive Ай бұрын

To do that you should use the tool agent combined with the retriever tool in Flowsie

@RolandoLopezNieto Ай бұрын

@@OnyxStudiosInteractive thanks, I'll try that today.

@khalidkifayat 6 ай бұрын

nice explanation. few questions 1. how/where to find clients 2. what should be the rough charge of this chatbot (excluding other expenses). Thanks in advance.

@OnyxStudiosInteractive 5 ай бұрын

For finding clients, you'll need to do daily prospecting and talking to people, presenting demos etc. As for expenses, token costs for Mistral are pretty inexpensive....input tokens cost 14 cents/ million while output tokens are 42 cents per million for the Mistral tiny model.

@khalidkifayat 5 ай бұрын

@@OnyxStudiosInteractiveappreciated

@bruceb85 Ай бұрын

@@OnyxStudiosInteractive there are dozens of already made tools out there already, like zapier ai chat, what would make this different? thanks

@martg0 6 ай бұрын

Great video! Do you host this in a docker and use the widget chat?

@OnyxStudiosInteractive 5 ай бұрын

Glad you enjoyed it! I am using docker to run the container for Flowise. The widget chat is pretty good. Although you can also use other widgets like VoiceFlow's chat widget for more interactive chatflows.

@JunaidAmjadPhotography 6 ай бұрын

Do we have to talk in keywords like calc requirements etc or can it process queries like "How much will it cost?"

@OnyxStudiosInteractive 6 ай бұрын

You use both however when it comes to calculations I prefer a custom function rather than rely on the LLM to do calculations. That way the results are more consistent.

@lalamax3d 3 ай бұрын

what to do if latest flowise Conversational Retrieval Agent is depriciated now...any ideas.??

@OnyxStudiosInteractive Ай бұрын

Well there's still the Conversational Agent that you can use. I think the only reason the Conversational Retrieval Agent is being deprecated is because it's basically the same as the Conversational Agent.

@florentflote 4 ай бұрын

@amrpyt 4 ай бұрын

Can i use any other chatbot instead of openai ?

@OnyxStudiosInteractive 4 ай бұрын

You can use other models. A lot of models support the OpenAI API or can also be found on OpenRouter. I use Mistral myself in a lot of my Chatflows. The only catch is that until recently OpenAI was the only LLM that did function calling. I think Mistral Large can do function calling now and maybe Claude as well.

@eatchocolate 5 ай бұрын

Why wouldn't you do the calculation right in flowise's JavaScript in the tool?

@OnyxStudiosInteractive 5 ай бұрын

You can definitely do that. Although I guess it depends on what you want to do. Outside of Flowise you can run automation workflows and so its helpful to be able to do calculations inside and outside of Flowise.