AI Transparency Policy
Table of Contents
Our commitment
We use AI in a few places to make Voipstudio faster and more useful. We also think you’re entitled to know when you’re talking to a machine, when a human’s behind the wheel, and what part AI played either way. This page explains exactly that, in plain language.
This policy reflects our obligations under Article 50 of the EU AI Act, but it goes further in places: we’d rather over-explain than leave you guessing.
AI in our support conversations
Some of our support replies are generated entirely by an AI system, trained on Voipstudio’s own manuals, documentation, and past support conversations, without a human reviewing that specific reply before it’s sent.
When that’s happening, we tell you at the start of the conversation, every time. No exceptions, no fine print.
You can always ask for a human. Say so at any point and we’ll bring one in. AI handles routine, well-documented questions well. It’s not the right tool for everything, and we’d rather hand you off than have it guess.
The same goes if you call us: some of our support calls are sometimes answered by an AI voice agent instead of a person. When that happens, it identifies itself as AI at the start of the call, and you can ask for a person just as easily as you can in a chat.
AI in our sales communications
We label sales messages based on how each one was actually made, not just which team it came from, because that’s what actually determines whether AI made the decision to reach out to you.
Messages sent without a person reviewing them first (for example, the opening message in some of our automated LinkedIn outreach) are written and sent entirely by an AI system, with no one on our team reviewing that specific message before it reached you. We treat this exactly like our AI-generated support replies: clearly labelled, no exceptions.
Messages a person reviewed and sent (the large majority of what our sales team sends, including most follow-ups) are written by a real person, sometimes starting from an AI-drafted first pass. The name on the message is someone who read it, could have changed it, and chose to send it as their own. The law doesn’t require us to flag this category, since a human made the call to send it, but we mention it anyway because we think it’s a reasonable thing to want to know regardless.
The same applies to phone calls: if an AI voice agent answers when you call our sales line, it says so right at the start, and you can ask for a person at any point.
AI in our blog and social media content
Much of our blog and social media content is entirely written by our own team, with no AI involved. Where we do use AI to help draft something, it’s never published as-is: someone on our team edits, adds to, and reviews it before it goes out, the same as everything else we publish under our name.
AI in the Voipstudio platform (for our business customers’ callers)
Separately from the above, Voipstudio sells AI-powered features, including call transcription and sentiment or CX analysis, to our customers, who are businesses using Voipstudio to run their own support and sales lines. Some of these businesses also connect their own AI voice agent, for example from Retell, to their Voipstudio account; that’s a product they choose and set up themselves, not something Voipstudio provides. If you’re calling one of those businesses rather than Voipstudio directly, any AI you encounter there is theirs to disclose, not ours, whether it’s a connected voice agent or one of our own analysis features running in the background. That said, we build our own features so they can meet that obligation properly:
- Where a business has connected its own AI voice agent through a provider like Retell, identifying it as AI to callers is between them and that provider, not something Voipstudio builds into our platform.
- Where a business uses our sentiment or CX-analysis features, we provide the tools for them to disclose that to their own callers.
If you’re unsure whether a company you’ve called uses Voipstudio’s AI features, that’s a question for them, not us. We’re the vendor, not the party on the call.
Using AI in our CX features
Some of our CX features use AI to look at what was said on a call, the transcript, rather than the recording itself. Depending on what’s switched on for your account, this can include automatic summaries, suggested follow-up actions, and CX Score, a quick way of surfacing how a call went for the customer.
These are optional, and you decide whether and how to use them.
If you use them to review your own team’s calls rather than just understanding your callers, that’s a different situation from everything else on this page, since it concerns you and your own staff rather than you and us. We’d suggest checking your own obligations around using AI on staff and customers, in whichever jurisdiction you operate in, so everyone is treated fairly and in line with the rules that apply to you. If it would help to understand more about how the technology works as part of that review, we’ve created a manual here.
How this looks in practice
The rule behind all of this is simple. If an AI system sends something to you without a person checking it first, we say so. If a person reviewed and sent it, even if AI helped draft it, we don’t have to flag that, though we’re happy to if you ask. And if AI started something and a person takes over partway through, for example a human joining a conversation an AI system began, that handover is usually made clear too.
The exact wording we use for any of this may change over time, but the rule doesn’t.
Questions
If anything here is unclear, or you want to know more about how a specific interaction was handled, contact us at dpa@voipstudio.com. We’ll give you a straight answer.
