The human-machine translator: CX's most important new role

The term “customer” can no longer be assumed to refer to a human, meaning CX must rethink everything it knows about customers, relationships, and the skills required to do the job. Katja Forbes explains

Add bookmark

Listen to this content

Audio conversion provided by OpenAI

Katja Forbes
02/26/2026

At 9:23am, CloudFlow Systems received two simultaneous inquiries for the same data storage solution.

ProcureIQ, an autonomous procurement agent, hit their API requesting technical specifications: 99.99% uptime SLAs, SOC 2 compliance, and sub-second response times for pricing requests. Anna, the CTO at ProcureIQ's parent company, called CloudFlow to discuss the strategic implications of migrating mission-critical client data.

By 11am, both had reached the same conclusion through completely different pathways: CloudFlow was the right choice.

CloudFlow won because they had mastered something most organizations have not yet named. They were operating as human-machine translators, and that capability is fast becoming the most important competitive advantage in customer experience.

Don't miss any news, updates or insider tips from CX Network by getting them delivered to your inbox. Sign up to our newsletter and join our community of experts.

The job description that just expired

CX has always operated on one foundational assumption: customers are human. We design experiences for humans navigating decisions with emotions, context, relationships, and non-linear thinking. We are very good at it.

That assumption no longer holds true.

When a single purchasing decision now routinely involves both a human executive and an autonomous procurement AI evaluating your organization simultaneously, you are orchestrating two parallel journeys that have to converge without conflicting. The skill set required to do that well is different from what got most of us here.

Jeff Gothelf, author of Lean UX and Sense and Respond, captures the transition clearly: "Ultimately, you could design the most amazing machine-to-machine experience. But if humans aren't willing to let go...we're going to need to design experiences that build in that human friction to make sure that people are comfortable transacting like this."

The implication for CX is direct. For the foreseeable future, we have to hold both tracks in mind at once.

What the dual track actually looks like

The practical starting point is content strategy, which is arguably the most critical CX function right now for organizations serious about machine customers.

In a parallel evaluation scenario, where a human and a machine assess your offering simultaneously using completely different criteria, a single-layer content approach simply does not work. What is needed is a multi-layer content strategy with three distinct components.

The surface layer is optimized for humans: visual storytelling, emotional value propositions, relationship-building touchpoints.

The data layer is optimized for machines: structured metadata, real-time APIs with performance metrics, machine-readable trust signals.

The integration layer is where both evaluations converge, the shared decision points where human and machine conclusions must align.

At every major customer touchpoint, the CX translator needs to ask three questions:

How does this serve machine evaluation criteria?
How does this serve human decision-making needs?
Where do these requirements conflict, and how do we satisfy both without compromise?

That third question is where most organizations currently have no answer.

The handoff problem nobody is designing for

The clean parallel evaluation scenario is only one version of reality. In practice, the interaction between human and machine customers looks more like an iterative loop.

The machine identifies five vendors meeting technical criteria. The human says the price is too high and adds sustainability requirements. The machine refines to three vendors. The human asks for more detail on one supplier. The machine flags a supply chain risk. The human asks the machine to reconsider the original five. And so on.

This is what the World Economic Forum describes as "constantly incorporating learnings from previous deployments to refine and improve the technical and organisational alignment of AI systems over time."

In customer experience terms, it means your organization has to perform well across multiple evaluation cycles you do not control, in three different directions: forward handoffs from AI to human, backward handoffs from human back to AI with new parameters, and circular handoffs where previously discarded options are reconsidered in new context.

Each of these handoffs is a CX moment. Each one is an opportunity to either support the decision-making process or introduce enough friction to trigger abandonment. Most organizations are designing for a single pass and will be caught flat-footed by the loop.

As Bruce Temkin, widely regarded as the godfather of CX, observes: “AI doesn't know what you meant, only what you told it to do. That gap between intention and instruction is where things go sideways.”

The translator role

What this requires of CX leaders is a genuinely new lens, one that holds machine evaluation logic and human decision-making needs in view simultaneously, and designs for both without defaulting to one or the other.

Don Scheibenreif from Gartner frames it as a balance that demands deliberate effort: "When we introduce more technology, we have to be as rigorous about introducing humanity into that...Organizations that can balance the two will be effective in the long run."

The CX leaders who develop fluency in both languages will find themselves in a different category from their peers. The translator role will become explicit in some organizations through new job titles and reporting lines. In others, it will remain embedded in existing CX leadership. Either way, the function will emerge, because the business need for it is already here.

The organizations that will define the next decade of customer experience are designing for both humans and machines simultaneously, without compromise. That is what the translator does. And right now, almost no one is doing it at scale.

You can find out more about Katja's book,
Machine Customers: The Evolution has Begun, via this link.