How to make the chatbot retrieve all answers from chat gpt, claude or any other llm without having to upload a knowledgebase

Sendbird AI Chatbot
1

Is it possible for the Sendbird AI Chatbot to retrieve answes from chat gpt, claude, gemini or other LLM’s with or with out having to upload a knowledge source.

just the way chat gpt, claude ai and gemini can virtually answer any question.

so that a user can ask any kind of question and get meaningful replies.

eg. a coding assistant or a story teller . etc

thanks

2

Hello Emmanuel,

It is your first Community post! Thanks for asking.

Let’s break down your question.

1.Is it possible for the Sendbird AI Chatbot to retrieve answes from chat gpt, claude, gemini or other LLM’s with or with out having to upload a knowledge source.

We support OpenAI GPT4o, GPT4-Turbo, Claude Sonnet 3.5 and other models such as Solar or LLama2. So yes you can leverage their base capabilities right off the bat. You may connect these models with a knowledge base to ground their responses and reply to the question.

  1. “just the way chat gpt, claude ai and gemini can virtually answer any question.” As you mention, these AI models (Large Language Models) may appear to answer any question. Of course, they don’t know all answers and can hallucinate. This is why you upload documents and URLs to our Knowledge Base from which the Ai chatbot (powered by LLMs) can retrieve relevant context and answer in an appropriate manner. This does not mean they answer correctly all the time but it lessens the risk of answering only on the LLM’s pre-trained knowledge. Knowledge stored in the base models are likely outdated and hence caution is needed when asking for latest information.

  2. “a coding assistant or a story teller . etc”
    These Large Language Models based Ai chatbots can be prompted or instructed to support your needs based on a particular need. You can “ask” a chatbot to role play as a coding assistant for Python, for example, or an expert story teller that uses language fit for a 5 year old.
    But all large language models hallucinate to a degree and hence caution is needed when using these chatbots. Careful evaluation of model responses, prompt instructions and user feedback is needed before any serious deployment.