Examples

Meticulous Researcher

Example of the agent using semantic scholar to retrieve papers.

The agent used Semantic Scholar to find papers and proceeded to rank its findings by relevance

LLM vs. Agentic AI

Agentic AI addresses the "knowledge gap" of LLMs by enabling the use of tools. Notice how the LLM alone produces a nonsensical response, but the same LLM given access to web search produces a correct response.

No tool access leads to a nonsense response (hallucination)

Allowing the agent to search the web at will leads to a correct response

Coding Assistant

Note how the custom "Coding Expert" prompt ensures the response includes notes on how to adapt to your own data. In this example, Qwen Code was able to generate a response without using any tools.

The LLM was able to generate a SQL query without using any tools. Custom prompt ensured the response included notes on how to adapt to your own data.

Multi-tool use

The AI Agent can use multiple tools as needed.

The AI Agent used a web search and a Wikipedia search to respond to the user's query

Multi-Agent chat

Kaimana allows you to chat with multiple LLMs in the samne conversation. In this example, we submit a query to DeepSeek R1, and then ask Qwen to add to R1's response.

Submit query to DeepSeek R1

Ask Qwen whether it has anything to add