Prepare and draft
Summaries, drafts, research, records updated. Low-risk work the agent does on its own.
Trust & Safety
Connecting your phone line, inboxes and CRM to AI agents is a serious decision. This page says what we do, what we don’t, and what we don’t claim.
Your data
Your data is used to run your agents, nothing else. It is never sold, never shared for marketing, and deleted on request at any time.
Control
Each agent has its own permissions. It only reads and does what its role needs.
Summaries, drafts, research, records updated. Low-risk work the agent does on its own.
Proposals, replies, posts and documents wait for a named person to approve them.
Pricing exceptions, complaints, sensitive or unclear cases go straight to a person, with the context.
Every action is recorded in an activity history you can review.
AI disclosure
Every call starts with IRIS introducing itself as an AI assistant, and the caller can ask for a person at any time. Callers are informed when a call is recorded. This is how we apply Article 50 of the EU AI Act, in force since 2 August 2026.
“Good afternoon, you’re speaking with IRIS, the AI assistant of Brightwell Facilities. I can help you or put you through to a colleague. How can I help?”Example greeting, adapted to each client
Providers
What your agents rely on, by purpose. The full list of subprocessors under the GDPR comes with your agreement.
The providers used for each purpose, and where they process data, are named in your data processing agreement. We tell you before adding a new one.
Documents & rights
We don’t claim certifications we don’t hold. These documents are available before anything is connected.
Questions
Anything that leaves the company waits for approval, so most mistakes are caught before they matter. Every action is recorded; we review exceptions with you every month and adjust the rules.
Yes. Each connection can be cut on your side or ours, immediately. When you stop, we close every connection and delete your data on request.
No. The shared business context holds what you approve, and each agent only reads the part its role needs. Sensitive information is used only when the process requires it.
Discovery call
Show us one process that slows your team down. We tell you honestly what the agents would do, what stays human, and whether it is worth building.