> ## Documentation Index
> Fetch the complete documentation index at: https://docs.diga.io/llms.txt
> Use this file to discover all available pages before exploring further.

# About Melo

> Why Melo can occasionally provide incorrect responses and how to use it effectively

# Melo can make mistakes

Melo is the AI assistant inside Diga's dashboard. It helps you build, test, and operate your voice agents — answering questions, suggesting changes, and guiding you through the platform.

Like any frontier AI assistant, Melo can occasionally produce responses that are **incorrect or misleading**. This is a known limitation of generative AI models — sometimes called **"hallucination"** — and it's worth keeping in mind whenever you act on Melo's output.

## Why this happens

A few situations where Melo's response may not be reliable:

* **Outdated information.** Melo may not have been trained on the latest changes to Diga, third-party integrations, or external services, and it can get confused when asked about very recent updates.
* **Plausible but inaccurate detail.** Melo can produce prompts, configuration values, or numbers that look authoritative but aren't grounded in your actual project.
* **Limited project context.** Even when Melo has access to your project state, it may miss context that lives outside the dashboard — your team's conventions, an external system's behavior, or a custom workflow.
* **Ambiguous prompts.** Vague questions tend to produce vague or speculative answers. The more specific your question, the more reliable Melo's reply will be.

## How to use Melo well

<Note>
  Treat Melo as a **knowledgeable collaborator**, not a singular source of truth. The final decision — and the responsibility for what runs in production — is yours.
</Note>

A few practical habits:

* **Verify high-stakes advice.** Anything that touches a live agent, billing, telephony routing, or a published workflow should be double-checked against the relevant page in this documentation or with your team.
* **Cross-check generated content.** If Melo writes a prompt, a tool definition, or a configuration value, read it before saving. A response that *looks* correct can still be subtly wrong.
* **Provide context.** Reference the agent, project, or feature you're working on. The richer the context, the better the response.

## Send feedback

If a particular response was unhelpful or wrong, let us know — your feedback helps us improve Melo over time.

* Use the **thumbs-down** button on any chat message to flag the issue.
* Or email us at [contact@diga.io](mailto:contact@diga.io) with details and suggestions.