The customer is going agentic

The platforms want to control the path between intention and outcome

Assistants are gradually becoming entry points to search, choice and action. OpenAI is turning ChatGPT into a space where the customer can refine their need, compare offers and connect with a merchant. Anthropic wants to let Claude act within the tools the user chooses, without turning the conversation into advertising space. Google connects Search, Gemini, Merchant Center and Google Pay to guide the customer from discovering an offer through to the transaction and its follow-up. Apple lets Siri draw on the user's personal context and directly trigger certain app functions. Meta is designing Muse to take in a goal, browse the web, fill in forms and ask for authorization before committing to an action. Shopify makes merchants' catalogs accessible to these agents. Stripe and Mastercard provide the payment, identification and authorization mechanisms they need in order to act within defined limits.

Adoption remains uneven. The strategic direction, however, is clear: the platforms are positioning themselves between the customer's intention and the outcome the company produces.

A brand can keep its products, its website and its apps while losing control of the path that leads to them.

This new distribution of power forces the company to define the place it wants to hold in relation to agents.

Faced with agents, the company must define its boundaries

These choices define the boundaries of the agentic relationship. They then set two requirements: being found and understood when the customer no longer starts their journey on the brand's own interfaces, and then letting the customer's agent act within the limits the company has chosen.

GEO makes the brand visible. Agent-to-agent makes it actionable

GEO helps an assistant find, understand and cite a company. It lets the brand be visible in generated answers.

The agentic customer also makes action requests: to book, change, buy, have delivered, repair, dispute or resolve. The company must then be found, understood and chosen, but also become actionable and keep its commitment.

The relationship becomes agent-to-agent when the customer's agent interacts with the company's agents, services and data to prepare an action, authorize it, carry it out and verify its result.

This relationship goes beyond a technical exchange between two pieces of software. It commits a promise, a power, a responsibility and a proof.

To become actionable, the company must therefore design the way it works differently. It can no longer impose a single journey on every customer. It must make available the reliable building blocks from which agents can compose their own path.

Faced with the mission entrusted by the customer, the company begins by defining the promise it wants to make executable. This promise specifies the outcome it chooses to take on and gives direction to the processes, capabilities and agents it mobilizes.

The company standardizes capabilities. The agent personalizes the path

The digital company designed standard journeys to serve large numbers of customers consistently: the same steps, the same interfaces and the same rules for moving forward. These journeys are the visible expression of standardized processes that organize in advance the way the company handles a request.

The agentic customer follows a different logic. Their agent can start elsewhere, choose a different order, combine several services and revise its plan when the situation changes.

The company then focuses its effort on what must stay stable: reliable data, explicit rules, accessible actions, controlled limits and a clear commitment.

A capability brings together the knowledge, data, decisions, operations, people and controls needed to contribute to an outcome in a repeatable way.

Each function must make part of its expertise accessible to agents. The business lines formalize the rules and the exceptions. Operations make visible the resources they can commit. Tech controls identities, rights and stop mechanisms. Human experts take over the situations that require their expertise, their empathy or their accountability.

The same combination of data, rules and resources can allow several valid answers. An insurer can direct its customer to the repairer able to act the fastest, let them choose their usual garage, or favor a solution with no out-of-pocket payment. These paths are all feasible, but they do not produce the same experience, the same brand difference or the same cost for the company.

Capabilities therefore answer a first question: what can we do? They are not enough to decide between several possible options. The company needs a guiding principle that indicates which outcome to favor, which customers it applies to and the conditions of that commitment.

This is the role of the promise. It answers a different question: what do we choose to make true, first and foremost, for the customer? It gives a shared direction to teams, systems and agents when several paths are possible.

An API makes a function accessible. A promise sets the priority outcome

In concrete terms, an API tells the agent which information it can access and which actions it can trigger. A process organizes how that function is carried out: the steps to follow, the rules to apply, the teams to mobilize and the checks to perform.

The API and the process make the action possible. On their own, they do not determine which outcome should take priority when several paths are feasible. The promise provides that direction. It specifies the outcome the company chooses to favor for a given type of customer, in a given situation, along with the particular way it intends to contribute to it.

The API

"what can we trigger?"

The process

"how will we carry it out?"

The promise

"which outcome do we want to make true, first and foremost, for the customer's agent?"

A promise connects three dimensions: the outcome the customer is seeking, the brand's difference and the economic value that makes the commitment sustainable.

An executable brand promise translates that ambition into decisions, capabilities, mandates and proof that teams, systems and agents can mobilize.

Keeping a promise does not mean controlling every cause of the final outcome. The company commits to the part of the outcome it controls, specifies what depends on outside players and remedies the gaps for which it is responsible.

Making this promise executable gives agents a framework to act within. That framework must also specify how far they can go, because the ability to act only creates value if the customer and the company can trust it.

Trust is built action by action

The customer does not grant their agent a blanket power. They can allow it to compare without letting it pay, accept a refundable purchase and require a confirmation before any lasting commitment.

Trust grows when the action stays understandable, when its consequences are proportionate and when a mistake can be detected and then put right.

The autonomy of agents follows the same logic. Each mandate specifies what an agent can do, on the basis of which facts, with which tools, within which limits and under whose responsibility. Proof then makes it possible to extend, reduce or withdraw that power.

Agentic governance does not slow action down. It makes it possible to act at the right speed, with a power proportionate to the consequences.

Taken separately, capabilities, mandates and integrations can pile up with no shared direction. They become a company-wide transformation when a single promise makes it possible to order them.

The Promise-Led Company organizes the business to keep its commitments

A Promise-Led Company responds to the mission entrusted by the customer by defining the promise it wants to make executable. It tests that promise against the reality of the company, then organizes its capabilities, its mandates and its proof to keep that commitment in an agent-to-agent relationship.

It does not measure its transformation by the number of models or agents deployed. It assesses it through a simple question: what outcome does the customer obtain today that the company could not produce end to end yesterday?

The agentic transformation starts with the customer and the mission they entrust. The company's transformation starts with the promise it chooses to make executable in order to answer it.