What the figures, benchmarks and calculator on this site do and do not mean, why none of it is legal advice, and where an AI voice agent stops. Written to be read before you rely on anything here, not after.
This site explains what our AI voice agents and call analytics do, quotes published research about phone service and sales, and offers a calculator for comparing costs. All of it is general information for evaluating whether to talk to us. None of it is a professional opinion about your business, your obligations or your numbers, and none of it is an offer that binds either side.
The company details page says who operates this site, the privacy policy says how we handle personal data, and the cookie policy lists the two cookies we use.
Whether you may call a given number, on what basis, at what hours, with what recording notice, and whether you must tell the person they are speaking to an AI, depends on your jurisdiction, your industry, how the number was obtained and what the person was told at the time. We write about those rules because buyers ask, and we name the instruments involved so you know what to look up. That is orientation, not advice, and it does not create a lawyer-client or consultant relationship.
Rules change and our pages carry a date, not a guarantee of currency. Before you launch a calling programme, get a written opinion from a qualified professional in each market you dial. Where we run an agent for you, the obligations that sit with you as the party doing the calling stay with you; the security and compliance page sets out which ones those are.
We separate three kinds of number, and we label them differently on purpose.
Statistics from named publishers and analysts are attributed in the sentence that uses them, with the year. They remain the property and the responsibility of whoever published them. We quote them as published; we do not restate them as our own findings, and a figure measured in one market does not transfer to another.
Some widely cited numbers are industry compilations, vendor-reported results or estimates rather than peer-reviewed studies. Where that is the case we say so in the same sentence as the number, because the hedge is part of the fact. Treat them as an order of magnitude, not a forecast of your result.
A few figures come from how our own systems and our founders' call-centre operations run — the share of calls a manual quality team can review by hand, the reply latency we design for, our published per-minute rates. These are our operating experience, stated as such. They are not research, they carry no publisher, and they are not a prediction of your numbers.
We publish no client case studies, no recordings or transcripts of real customer calls, and no proprietary research, because we have produced none. If you ever see one of those attributed to us, it is not from us. Where a claim on this site cannot be sourced, we argue it qualitatively instead of inventing a number.
The cost calculator compares an in-house team, an outsourced call centre and an AI agent using assumptions we state on the page: minutes per call, occupancy, staffing overhead and our published per-minute rates. It is arithmetic on inputs you choose, run in your browser. It does not know your salaries, your local employer costs, your telephony contract, your call mix or your conversion rate.
Use it to see which direction the numbers point, not as a budget. Nothing it displays is an offer, a quotation or a commitment on price. Published rates exclude VAT and any other applicable tax, are set per region, and change with volume; the only figures that bind us are the ones in a written proposal signed for your project.
We would rather set the limits out here than let a demo imply there are none.
The recordings on our examples page are scenarios we produced to show voice quality, interruption handling and script logic. They are demonstrations, not recordings of identified customers, and they are not evidence of a result you will get. A demonstration call placed to a number you enter runs the same way: it shows behaviour, not outcomes.
Our roadmap and any page describing what a deployment looks like over time set out intentions and a recommended order of work. They are not commitments to deliver a particular feature, on a particular date, or to reach a particular result. Where we do not know how long something takes, we say so rather than quote a number of days we cannot hold to. Anything we do commit to for your project belongs in a signed document, not on a web page.
We keep this site accurate and available, but we do not warrant that it is free of error or interruption, that every page is up to date, or that it fits a particular purpose of yours. Pages are corrected as we find problems, and legal pages carry the date of their last change.
Links to other sites, and the one external service this site calls, are outside our control: we do not endorse their content and we are not responsible for their practices. Which external service is involved, and when, is listed in the cookie policy.
To the extent permitted by applicable law, we are not liable for decisions taken, or losses incurred, in reliance on the general information published on this site — as distinct from obligations we accept in a signed agreement with you, which this page does not limit. Nothing here excludes liability that cannot lawfully be excluded, and nothing here restricts your rights under mandatory consumer or data-protection law, including the rights described in the privacy policy.
This site is published in English, French and Spanish. The translations are provided so you can read it in your own language; if a difference in meaning ever arises between versions, the English text is the reference one.
If a figure, a claim or a limit on this site is unclear — or looks wrong to you — write to hello@benerra.ai and tell us which page. We would rather correct it than defend it.
