Synthetic Users

Acting like users by prompting takes ages and is not a viable long-term solution, especially when it comes to agents. We're changing that. Finetune empowers developers to create detailed synthetic users that reflect their customers. This way, when you enter a fine-tuning session, you can run your agent against these users.

Minimal setup

How do we create these synthetic users? All you have to do is provide us with a link to your website's home/about page. Our in-house multi-agent system, powered by fine-tuned models and relevant census data, can create detailed personas equipped with a set of questions and a mission statement.

Users' structure

What do these users look like? Here's a simple JSON object to reflect the users:

{
  "Name": "Luca Müller",
  "Age": 30,
  "Questions": [
    "What is the process for getting a ticket refunded?",
    "How long does it take to receive a refund?",
    "What documents are needed to apply for a refund?"
  ],
  "MissionStatement": "I want accurate information on how to get my ticket refunded."
}