> For the complete documentation index, see [llms.txt](https://docs.hivel.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.hivel.ai/configurations/enabling-hivels-ai-powered-features.md).

# Enabling Hivel's AI-Powered Features

Hivel’s AI-powered features unlock deeper insights and surface meaningful patterns across your software development lifecycle, like Comment Categorization, Code Telemtry, Lucy (AI Insights Chatbot), etc.

{% hint style="success" %}
**AI Features Are Opt-In Only**

By default, all AI-powered features in Hivel are **disabled**. These features are only activated:

* Upon **explicit customer consent**, or
* When a user **enables them manually** in the settings.
  {% endhint %}

### How to Enable or Disable AI Features

<div data-with-frame="true"><figure><img src="/files/oPi71VnpUbaJMDhm2mlo" alt=""><figcaption></figcaption></figure></div>

#### Steps:

1. Go to the **Configuration** screen.
2. Scroll to the section labeled **AI**.
3. You can choose to enable all or few of the features.

> ⚠️ Turning the toggle ON allows Hivel to process select metadata via large language models (LLMs) to generate insights.

<div data-with-frame="true"><figure><img src="/files/lJb5rvH3YDgs53oBP9YG" alt=""><figcaption></figcaption></figure></div>

### Data Privacy and Security Commitments

<h4 align="center"><a href="https://www.hivel.ai/privacy-policy#:~:text=AI%20Model%20Engagement%20and%20User%20Data%20Autonomy"><strong>AI Terms and Conditions</strong></a></h4>

Hivel prioritizes **data security, anonymity, and customer trust**. Here's how we protect your information:

#### **What Data Is Shared**

Only the minimum required data is shared, and it's processed with care:

* **Anonymized data only**: No company names, author identifiers, commit hashes, ticket numbers, or branch names are sent to the AI model.
* **Aggregated content**: We batch similar types of data (e.g., commit messages or PR comments) and remove any contextual identifiers before sending.
* **Zero business logic exposure**: We do not send your source code, configuration files, or deployment secrets.
* **No access to customer environments**: AI models interact only with sanitized metadata already available within Hivel’s controlled systems.

#### **Ephemeral Processing and No Retention**

* Data is transmitted via secure APIs using **TLS encryption**.
* AI model providers do **not store** any of your inputs or outputs.
* Every interaction is **stateless** i.e. each input is processed in isolation and discarded immediately after generating the response.

### **FAQs**

#### **Is any customer code shared with AI models?**

No. We do **not** share source code, internal logic, or proprietary configurations. AI features operate only on derived metadata (like commit messages or review comments), after rigorous redaction and obfuscation.

#### **Which AI models are used?**

We use industry-grade LLMs provided by OpenAI and Anthropic, interfaced via private, stateless API calls. These providers are contractually and technically prohibited from retaining or training on your data.