Robin uses AI in 2 ways:
- Robin's own AI features. Robin builds and runs these features inside Robin. They use Google's Gemini models.
- AI assistants your employees connect to Robin, such as Claude and ChatGPT. Your organization or your employees choose these assistants, and their providers handle the data under their own terms. Administrators control these connections in Robin.
This article covers both.
Robin's AI features
Robin's AI features use Google's Gemini models:
| Feature | What it does | Data it uses |
|---|---|---|
| Robin user agent | Suggests desk and resource bookings for the rest of the week, or for the rest of this week and next week. The user accepts each suggestion before Robin books it. | The user's reservations from the past 4 weeks, their default office and working hours, and availability |
| AI assistant for Advanced Analytics | Answers questions about desk and space use | Your organization's past desk and space use, and its people inventory |
| Meeting room assistant | Resolves booking conflicts in a meeting room when the user asks. It suggests other rooms of a similar size and on similar floors for the conflicting meetings. It does not change meeting times. Nothing changes until the user accepts a suggestion. | The room's bookings, the size, location and availability of other rooms, and the user's calendar permissions |
Robin's AI features run only when a user asks. They do not process your organization's booking history in the background.
Organizations hosted on Robin's EU infrastructure can use the Robin user agent. They do not have the AI assistant for Advanced Analytics or the meeting room assistant.
The Robin user agent and the meeting room assistant act as the signed-in user, within that user's Robin permissions. A user can accept a room change only for a meeting they are allowed to edit.
Employee data in the AI assistant for Advanced Analytics
The AI assistant for Advanced Analytics answers from your organization's Advanced Analytics data. This data includes employee names, email addresses, departments, groups, job titles, managers and default buildings. It also includes desk and room reservations, such as who booked, who was assigned and meeting titles. Titles of meetings marked private are excluded.
To answer a question, the assistant sends Gemini a description of this data and the results of the query it runs. The results can include employee names and email addresses. The assistant can answer questions about individual employees, such as how many desk reservations a named person made. It can also answer with totals for teams, buildings and the whole organization. It uses data from your whole organization.
The assistant can make mistakes. With each answer, it explains in plain language how it found the result. Check important figures in your Advanced Analytics reports before you act on them.
How Robin protects your data
Robin protects the data its AI features use in these ways:
- Robin does not train AI models on your data. Robin does not use your data to train or fine-tune any AI model.
- Google is the only AI provider for Robin's AI features. Robin's AI features use Google's Gemini models through Google Cloud's Vertex AI. Google is a Robin subprocessor. Robin's AI features do not send your data to any other AI provider. AI assistants your employees connect work differently, as the next section explains.
- Google does not train on your data. Under Google Cloud's terms for Vertex AI, Google does not use data that Robin sends to Gemini to train its models.
- What Google keeps. Google may cache data that Robin sends to Gemini, to make responses faster. Google deletes cached data within 24 hours. Robin does not turn on Google's optional logging of requests and responses.
- Data location. For organizations hosted on Robin's US infrastructure, Gemini processes requests in the US, and Robin stores AI conversations in Google Cloud's us-east4 region. For organizations hosted on Robin's EU infrastructure, Gemini processes requests in the EU, and Robin stores AI conversations in Google Cloud's europe-west3 region. Robin's AI monitoring tool keeps a copy of conversations in the same region, to monitor and troubleshoot the features.
- Conversation retention. Robin keeps conversations with the Robin user agent, the AI assistant for Advanced Analytics and the meeting room assistant for 90 days, then deletes them. A conversation includes the user's messages, the assistant's answers and the Robin data the assistant used. There is no setting to opt out of this retention.
- Controls. Administrators cannot turn off Robin's AI features, including the AI assistant for Advanced Analytics. The AI assistant for Advanced Analytics is available only to users with analytics access, in organizations with Advanced Analytics. The Robin user agent and the meeting room assistant change nothing until the user accepts a suggestion.
- Changes made with AI. A booking or room change made through Robin's AI features is an ordinary change by the user who accepted it. Robin does not mark it as made with AI.
AI assistants your employees connect
Employees can connect Robin to AI assistants such as Claude and ChatGPT through Robin's MCP server. Organizations hosted on Robin's EU infrastructure cannot connect AI assistants. This works differently from Robin's own features:
- Your organization or your employees choose the assistant. Robin does not send your data to these providers on its own. When an employee uses a connected assistant, the assistant requests Robin data on that employee's behalf, limited by their Robin permissions.
- The AI provider is not a Robin subprocessor. It handles your data under the agreement for the employee's AI account, not under Robin's agreements.
- Administrators control access. AI assistant connections are on by default. You can turn them off for your whole organization, and organizations with custom roles can limit which users can connect.
Robin cannot tell a personal AI account from an account your organization manages. A personal account usually falls under the provider's consumer terms. Robin recommends that employees connect only through AI accounts your organization manages. To enforce this, limit the AI assistants permission to approved users, or turn off connections.
To learn what data these connections can access, how to control them and what to check in a security review, see Security and governance for AI assistant connections. For setup steps for each assistant, see the Robin MCP site. For example prompts, see the Robin MCP prompt library.
Privacy and data protection agreements
See Robin's Privacy Policy for Robin's privacy practices. To request data protection and data transfer agreements, contact gdpr@robinpowered.com. To ask Robin to access or delete personal data, including AI conversations, contact privacy@robinpowered.com.