Artificial intelligence in Raynet helps you write and summarize texts, evaluate Leads, or chat with an assistant. To make this work, part of your data is processed by the provider of the AI model, that is, the service that actually generates the texts. This article explains what data leaves Raynet for each feature, who processes it, and what remains your responsibility.
The article is aimed mainly at administrators and at the people in your company who handle personal data protection. Raynet AI features are available in the PROFESSIONAL, ULTIMATE and ENTERPRISE pricing plans. AI Lead Scoring is still in Beta.
Which AI features Raynet offers
| Feature | What it is for |
|---|---|
| Raynet AI in text fields | Summary, text improvement, spelling correction, drafting an email or a reply, translation. |
| Account history summary | An overview of your relationship with the Account on its Timeline. |
| AI Lead Scoring | Evaluation of a Lead against your ideal customer profile. |
| Signature scanner | Recognition of contact details from an email signature. |
| AI Assistant | A chat that answers questions about working with Raynet. |
| MCP server | A connection between Raynet and an AI tool you already use, for example ChatGPT or Claude. |
Raynet AI in text fields and the Account summary never send anything on their own. Data leaves only when the user chooses an action.
Who we send data to
We use language models from Microsoft (Azure OpenAI) and Google (Vertex AI). Both are listed in our Subprocessor list and Data Processing Agreement, which you can find in the Artificial intelligence usage disclaimer dialog. They are listed the same way in the standard contracts and documents published on the Raynet website. For more information about how personal data is processed, see the data processing agreements of these companies.
What data is sent
For each feature we send only what it needs to do its job:
content you enter or select in the feature yourself, for example text in a field, an email, or details of an Account or a Lead,
your choices and settings, for example the language, the tone of the output, or the ideal customer profile.
Features you start with a click send data only at that moment. With the MCP server, the AI tool receives only the data it asks for.
Consent and control over the features
When a user first uses a Raynet AI feature, the Artificial intelligence usage disclaimer appears. Each user accepts it once for themselves by clicking Accept. An administrator can accept it on behalf of all users.
AI Lead Scoring is switched on by an administrator in Settings » Record settings » Lead » AI Lead Scoring. The administrator accepts the disclaimer, and it applies to the whole instance. Until the feature is switched on, no Lead is evaluated.
MCP keys are created by an administrator in Settings » MCP keys. For each key you choose the permission Read or Read and write. If you delete a key, the connected AI tool stops working.
Permissions and access to data
AI in text fields works with the text and the record the user currently has open. The Account history summary is compiled with respect to the user's access rights.
The MCP server enforces the standard Raynet permission model for every request. A key is also always bound to a specific user account. If the account has limited rights, the AI cannot reach data beyond those rights.
A key with the Read permission cannot change or create records, and the AI is not even offered the write tools in that case.
For MCP, we recommend choosing an account with the minimum necessary rights and creating a separate key for each AI tool and each user. If a key leaks, delete it and create a new one. You cannot restrict a key retroactively. Never paste a key into a chat with an AI.
What remains your responsibility
Checking the output. AI can generate inaccurate information, so always check important details. Raynet is not responsible for generated content.
Personal data. By using Raynet AI, you are responsible for the data you pass to the AI for processing. Consider how personal data processing toward your own clients is set up in your company, and add information for them if needed.
What you put into AI. For text fields, there is a simple rule: if you do not want something sent to the provider, leave it out of the field or do not use the feature on that text.
Third-party tools with MCP. We are not responsible for the use of AI tools that you connect to Raynet yourself. For sensitive data, we recommend choosing tools that meet GDPR requirements and let you configure data protection.
AI does not overwrite your original text. You can edit the suggestion, use it, or discard it.
When something doesn't work
| Message | Possible cause |
|---|---|
| Raynet AI is disabled | The feature is not enabled for your account. Get in touch and we will find out why. |
| Raynet AI is currently processing a high volume of requests. Try again in a moment. | The service is temporarily busy. Try the action again in a moment. |
| You've reached the query limit for the AI assistant. Please try again in a little while. | You have hit the query limit. The message tells you how long to wait before trying again. |
If you have more questions about data security when working with AI, contact us at podpora@raynet.cz. We are happy to help, or to pass your question on to the colleagues responsible for personal data protection.