Content validation and correction

AIsuru gives you a powerful tool to continuously improve your Agent's knowledge by validating and correcting AI-generated content.

Why is validating responses important?

Validating responses gives your Agent a broader, richer knowledge base: validated contents become part of your Agent's knowledge!

Adding new contents or question variants helps your Agent more easily find the right answer to your users' questions, improving their experience.

Where do I validate responses?

1. Go to your Agent's management panel;

2. In the sidebar, click on Contents;

3. Click on the "AI-Generated" tab.

What can I do there?

In the table, you'll find all the questions users have asked that were answered by the generative AI. You'll see several buttons:

Button	Description	When to use it
Find in conversation	Opens the conversation where the question was asked	To better understand the context in which a question was asked
Confirm	Validates the question and the answer provided by the generative AI	When all information is correct: question, answer, required topics, and role assignment
Add as question variant	Adds the user's question as a variant of an existing content	If you already have a content that answers the user's question, but the Agent responded poorly
Edit	Opens a panel where you can fully edit the content before validating it	If you want to change the question, the answer, or add customizations before validating
Delete	Removes the content from the table	When there's no correct answer for a specific question, or the question isn't useful

Note: AI-generated responses longer than 10,000 characters are not saved in this table.

Automatic context setting based on URL

AIsuru automatically remembers which page (URL) a question was asked from and sets it as a required topic. This lets the Agent give different responses depending on which page the user was on when they interacted with it.

For each content in the "AI-Generated" section, you'll find two required topics set automatically:

1. ROUTE:[AGENT_NAME];

2. PATHNAME:[PATH_LINK].

These topics let the Agent contextualize its responses based on where the question came from. If you want to disable this feature, you'll need to remove both required topics when validating or editing the content.

Don't worry if you forget to remove the required topics: you'll get a warning before saving letting you know they're there.

Duplicate contents

It's important that your Agent's contents aren't duplicated or contradictory: your contents should be clear and non-redundant.

Final thoughts

Validating AI-generated responses is a crucial process for continuously improving your Agent's performance. By spending a little time on this regularly, you'll not only grow your assistant's knowledge base but also sharpen its accuracy and relevance. Pay attention to the automatic URL-based context: it enables more targeted responses, but can be disabled if it's not needed for your specific use case. Make the most of this tool to build an Agent that's always improving and aligned with your users' needs.