Adding Real-Time Web Search and Local DB Search to OpenAI Assistant API

[empty wharf]

I currently have a chatbot built on the Assistant API that handles document uploads. Now, I would like to add two features:

Real-time web search when the answer is not available in the documents.
Local database search based on user requirements.
How can I implement these features?

[inner totem]

https://platform.openai.com/docs/assistants/tools/function-calling

[empty wharf]

I tried this but failed if you have xperience can you help me?

[inner totem]

there are many examples for function calling in openai cookbook and github. what was the issue?

[empty wharf]

i wrote like this but is not doing real search

{
"name": "google_for_answers",
"description": "Search Google using a fully-formed HTTP URL to retrieve relevant information and enhance knowledge.",
"strict": false,
"parameters": {
"type": "object",
"properties": {
"url": {
"type": "string",
"description": "The fully-formed HTTP URL to perform a search.",
"default": "https://google.serper.dev/search"
}
},
"required": [
"url"
]
}
}

[inner totem]

that's not how functions work. see the official detailed document https://platform.openai.com/docs/guides/function-calling

[empty wharf]

okay , if i succed to run the function and after that it will be possible to do real time web search

[inner totem]

you run the search on your server and only pass the results. you dont give any url, etc. to it

[empty wharf]

when you mean server do you mean platform.open.ai? Because i have already uploaded documents, just i need to add real time web search if the answers are not in the documents

[inner totem]

your own server that does all the api calls. openai servers will not run the search

[empty wharf]

thank you for this anwer , one more question from platform ai may i coonect to the local db? and gettin answer from local db

[inner totem]

you can make a function and make it get the data but the api cant connect directly