EY’s Future Consumer 2026 argues that whether people let AI agents act on their behalf will depend on trust and on consent they can withdraw. Outbound customer contact has always run on that principle, because the person you call never asked to be called, and the call is worth something only for as long as they choose to stay on the line.

This is an analysis piece built on a consumer survey, and it is not about telephone service. We read it for what it says, take care not to stretch it, and draw out what it means for banks, insurers and other regulated firms that contact their customers by phone.

What EY published#

“Future Consumer 2026: In the permission economy, who earns the right to act?” was published by EY on 19 June 2026, written by Laurence Buchanan. Its central thesis is that delegating actions to AI agents depends on trust and on consent that the consumer can revoke. As AI moves from answering questions to doing things, booking, paying, switching, cancelling, the question shifts from what a system can do to what a person has allowed it to do.

None of EY’s survey figures concerns telephone contact, and the temptation is to borrow numbers from elsewhere and attach them to the essay. We will not do that. Where we cite consumer data below, it comes from other publishers, named with their own samples and dates.

Outbound has always been a permission business#

Inbound contact starts with the customer’s intent: they called, so they want something. Outbound starts with the company’s intent. The customer did not choose the moment, may not recognise the number and may not want to talk. Every second of the call has to earn the next one.

Clients often assume outbound is simpler than inbound. Someone who calls you will cope even with an IVR. Someone you call may not even want to pick up, let alone keep talking. – Dawid Wójcicki, Founder & CEO of Vee

We measure that permission directly. The second rung of our KPI ladder, after reach, is continuation: the inverse of hang-ups after the customer hears the company name. It is the most honest measure of whether a customer has granted, for now, the right to continue. It is also revocable at any moment, which is what EY’s framework would predict.

The scale of the gap between trying and talking shows how conditional that permission is. Up to 31 October 2026, Vee made 298,684,764 call attempts and held 38,325,799 connected AI conversations, both cumulative Vee operating data. These are different units, single dial attempts and connected conversations, and they should not be turned into a rate, but the distance between them is the space in which permission is either granted or withheld.

Trust starts with saying who is calling#

If consumers will delegate only to systems they trust, the first condition is that they know what they are dealing with. In 2016 we said publicly that we would not hide from customers that they were talking to a machine, and we have kept to it. Today the obligation is also law: Article 50 of the AI Act requires transparency when people interact with AI, and those obligations have applied since 2 August 2026, with Commission guidelines published on 20 July 2026.

Other research suggests consumers care. Ping Identity’s 2026 global consumer survey, conducted by Talker Research among 11,000 consumers in 12 countries, found that 56% say knowing AI is being used matters more than simply getting the fastest result, and that 71% are concerned an AI agent acting on their behalf could misrepresent their personality or values. That second figure is about agents acting for consumers, which is the territory EY describes; it is a reminder that trust in delegation is not yet won.

Permission in outbound contact: what it looks like in a call
Principle in EY’s framingWhat it means in an outbound call
Trust before actionDisclose AI at the start; verify identity before any account detail
Clear scope of actionA defined purpose for the call; nothing outside it
Revocable consentThe customer can ask for a person, stop the call or withdraw consent to recording
AccountabilityEvery call reconstructable from recording, logs, statuses and script version

Left column paraphrases EY’s thesis; right column describes Vee’s operating practice in regulated processes.

Scope: an agent may only do what it was permitted to do#

EY’s question, who earns the right to act, has a direct counterpart in regulated contact: what is this call permitted to do? In collections, identity is verified before any detail of a debt is mentioned, as a rule built into the platform rather than a line in a script. Replies come from a closed, versioned library approved by the client. The conversation does not improvise outside the purpose it was built for.

That narrowness is sometimes described as a limitation. We see it as the condition for being allowed to act at all. A bank can let an automated call agree a payment date precisely because it knows the call cannot do anything else.

Consent that can be withdrawn only on paper is not revocable in any useful sense. In our processes the exits are designed in advance and agreed with the client: a customer who asks for a person, a failed identity check, withdrawn consent to recording, a death or legal dispute, a request for erasure, or a customer in financial hardship. Each leads to a transfer to the client’s own staff, a stop or a referral, not to a script that tries again.

Consent has a history in our work. In 2018, as GDPR took effect, we built a process specifically to collect GDPR consents by phone, with free conversation followed by an explicit yes-or-no confirmation. The lesson then is the same as EY’s now: the customer’s permission must be clear, specific and recorded.

Permission is also about when and how often#

Trust is not only built inside the conversation. It is built, or spent, by the pattern of calls around it. A customer who is rung at an awkward hour, or five times in a day from numbers they do not recognise, has already decided how the next call will go before anyone speaks. That is why we treat contact strategy as part of the customer experience rather than a dialler setting.

In practice that means market-specific call windows, attempt intensity and spacing that depend on what happened on the previous attempt, an outbound number strategy, and SMS before or after the call so the customer knows who is trying to reach them and why. A callback the customer asked for, a voicemail and a busy line are different signals and deserve different retry clocks. Each of these choices respects the customer’s time, and each is measured against reach and continuation rather than against volume of attempts.

The principle is the one EY describes, applied before the call starts: earn the right to the next contact rather than assume it.

Where the analogy stops#

EY is writing mainly about consumers delegating to their own AI agents, and about which brands those agents will be allowed to act with. An outbound call from a bank is a different situation: the company is acting, not the consumer’s agent. We should not claim the essay as evidence about telephone contact. What transfers is the principle, not the findings.

There is one development worth watching. Forrester predicts that around 10% of consumers may use a personal AI agent as their main service interface in 2027, with more bot-to-bot traffic. It is a prediction, not a measurement, but if it materialises, outbound calls will increasingly meet software acting with the consumer’s delegated permission. Scope, disclosure and revocability will matter on both sides of the line.

What to ask of your outbound programme#

  • Does every call say what it is? Disclosure of AI at the start, without exception.
  • Is the scope written down? One purpose per call, with replies from an approved library.
  • Can the customer leave easily? Map the exits: a person, a stop, withdrawal of recording consent, erasure, hardship.
  • Do you measure continuation? Hang-ups after the company name are the clearest signal of permission refused.
  • Can you prove what happened? Reconstruction of any call for your auditors and regulators.

EY’s permission economy is a useful frame for the next stage of AI in customer relationships. For outbound contact it is not new; it is the rule the work has always obeyed. The firms that keep consumers’ permission will be those that disclose, stay within scope and make leaving easy, and those are the terms on which we run calls for our clients.

Follow The First Second

Occasional issues on the mechanics and economics of AI-powered customer contact. No noise, no vendor news.

Follow via RSS