Forrester’s predictions for 2027 describe a year in which some consumers send AI agents to deal with companies, some firms hand whole customer journeys to agentic orchestration, and rising AI costs force others to cut back. Taken together they make one argument: AI in customer contact will be judged on measured results and cost, and human connection will become the part that differentiates.
Every figure below is a prediction, not a measurement of what is happening today. We set out what each publication says, mark its scope, and then explain what it means for regulated customer contact and how it compares with what we see in production.
Three publications, three predictions#
| Publication | Date | Prediction | Scope |
|---|---|---|---|
| Predictions 2027: Human connection will be the next battleground for customer success and service | 30 Oct 2026 | Around 10% of consumers may use a personal AI agent as their main service interface in 2027; bot-to-bot traffic will rise | Customer service and success |
| Predictions 2027: CX teams build new muscles to seize the AI opportunity | 21 Oct 2026 | One in five companies will implement fully agentic journey orchestration in 2027; one in four CX teams will use AI to transform legacy measurement | CX teams |
| B2C marketing, CX and digital predictions: rising AI costs will force B2C leaders to scale back a third of AI initiatives | 21 Oct 2026 | About one third of B2C AI initiatives scaled back because of rising costs | Mainly B2C marketing, not contact centres as a whole |
Forrester Predictions 2027, published as blog posts and a press release; full reports are for Forrester clients. All three are forecasts.
Human connection as the battleground#
Forrester’s first prediction has two parts. Around one consumer in ten may use a personal AI agent as their main interface for customer service in 2027, and more interactions will run between bots on both sides, especially in B2B, where Forrester sees the most fertile ground. Its headline conclusion is that human connection will become the competitive battleground for customer success and service.
This matches a view our founder has stated publicly more than once: in a January 2022 interview with Business Insider Polska and again on Radio Gdańsk in September 2025, he argued that most everyday interactions would come to be handled by AI and that human contact would become a premium good. We note the overlap; we do not claim the forecast as proof.
For regulated enterprises the practical meaning is not that people should handle more calls, but that the calls people handle must be the right ones. In our processes the route to a person is designed, not improvised: a customer who asks for someone, a hardship case, a dispute, a failed verification, each has an agreed path to the client’s own staff. Repetition goes to the machine; judgement and empathy go to people.
Human connection is also on our side of the service, not only the client’s. Vee employs no call agents, yet people are present in every project: dialogue designers, a voice studio, analysts and a quality team that listens to selected calls every day, especially the unusual ones. What customers experience as a well-judged conversation is the product of people listening to thousands of real calls and deciding what should change.
Bot-to-bot traffic#
If Forrester is right, outbound calls will increasingly be answered, or screened, by software acting for the customer. That raises practical questions about disclosure, verification and scope on both sides. We would not overstate how soon this will touch regulated telephone contact. Identity checks before any account detail, AI disclosure under Article 50 of the AI Act and a closed scope for each call are, however, the right foundations whichever kind of counterpart picks up.
New muscles: orchestration and measurement#
Forrester’s second prediction is that one company in five will implement fully agentic orchestration of customer journeys in 2027, and that one CX team in four will use AI to transform legacy CX measurement. The second half is the more certain and, for most readers, the more important.
Orchestration across a journey is a direction we are working in ourselves, and parts of it run today, coordinating calls, text messages and retry scheduling within one process. We describe it as a development direction, not a finished product, and we would treat any claim of fully agentic orchestration in a regulated process with care until it can show its controls.
Measurement is where the muscle is really built. We work with a four-rung ladder: reach, continuation, obtaining the key information, and the real business effect such as a payment or an attended visit. Cases carry statuses on four axes, technical, dialogue, processing and business, so the client can act on them daily. And results are judged against a comparator on the same buckets and portfolios, so the effect claimed is the incremental one.
The measurement shift Forrester expects is healthy, provided it pushes towards outcomes. A promise to pay is not a payment, and a booked visit is not an attended one. Teams that report only declarations will find their numbers look good in the first month and hard to defend in the sixth, when finance compares them with what actually arrived.
Rising AI costs and the third that gets cut#
The third prediction is the one most likely to be misquoted. Forrester expects rising AI costs to force B2C marketing leaders to scale back about a third of their AI initiatives. The scope is mainly B2C marketing. It is not a forecast that a third of contact-centre AI will be abandoned, and it should not be presented as one.
The underlying point does carry over. AI is not free to run, and its cost does not automatically fall. Gartner has made a related forecast for service specifically: that the cost per resolution of generative AI in customer service will exceed $3 by 2030, above the cost of many offshore human agents. Both are forecasts. Together they argue for choosing technology by the cost of a result rather than by its novelty.
That is how we choose. Our dialogue engine runs in several variants, from fully deterministic with no language model, through small models used only to classify intent, to larger models used for classification without generating replies. In regulated processes we use the lightest variant that does the job, because it is cheaper, more predictable and easier to audit. Deterministic where it matters, generative where it helps.
If we are paid by the minute, we are not playing towards the same goal as our client. – Dawid Wójcicki, Founder & CEO of Vee
Cost discipline also shows in how a service is billed. New Vee agreements are no longer billed by the minute; a minute rewards long conversations, while results often come from short ones. Since September 2026 we have also stopped regular public reporting of minutes and case volumes, in favour of reporting on economic value, effectiveness and process economics.
What the three predictions add up to#
Read together, the predictions describe a market moving from adoption to accountability. More AI will be in use, some of it acting for consumers and some orchestrating journeys for firms. More of it will be cut where it does not pay. And the human part of service will matter more, not less, because it is where the difference is felt.
None of this is certain, and Forrester does not claim it is. For a regulated enterprise planning 2027, the safest use of the predictions is as a checklist of risks rather than a forecast to bet on.
Questions for 2027 planning#
- Where must a person be available? Define the human route before automating, and measure how often it is used.
- What will a call do if software answers? Decide how identity, disclosure and scope apply when the counterpart is an agent.
- Can you measure the outcome, not the activity? Agree the KPI ladder, the comparator and the source of truth now.
- What is the cost of a single success? Compare options on cost per result, not per minute or per token.
- Which initiatives would you cut first? If AI costs rise, know which processes pay back and which only look busy.
Forrester’s predictions will be cited often in the coming year. Cite them as predictions, keep their scope, and use them to ask harder questions of your own programme and your suppliers. Those questions, about measured outcomes, cost per success and the place of people, are the ones we work through with our clients every week.
Follow The First Second
Occasional issues on the mechanics and economics of AI-powered customer contact. No noise, no vendor news.
Follow via RSS