AI agents don’t change commerce — they change who makes the decision

enero 30, 2026

Why commerce is no longer designed for people — and what leaders must do about it

In October 2025, McKinsey published a sharp analysis on what it calls agentic commerce: a model where AI agents don’t just assist consumers, but act on their behalf.

This is not another UX evolution.
It is a reallocation of agency.

When software starts making decisions — not just recommendations — the foundations of commerce change: discovery, differentiation, loyalty, payments, and trust. What follows is our synthesis of the most relevant insights from the report, and what they mean for organizations designing the next generation of digital experiences.

What Is Agentic Commerce — Really?

Agentic commerce describes a new paradigm in which AI agents autonomously execute commercial decisions for users.

These agents:

Understand intent, preferences, and constraints
Navigate digital commerce environments (sites, APIs, messaging)
Execute multistep actions — search, comparison, negotiation, purchase
Coordinate with other agents and systems without human intervention

This is not automation in the traditional sense.
It is delegation.

The consumer no longer shops — they authorize.

Why This Is a Structural Shift (Not an Incremental One)

Most e-commerce innovation over the last 20 years optimized for:

Faster checkout
Better personalization
Fewer clicks

Agentic commerce breaks that logic entirely.

When agents become the primary interface:

Clicks lose meaning
Funnels collapse
Interfaces are no longer designed for humans

Commerce stops being a journey and becomes a background process.

The Market Opportunity: Why This Matters Now

The scale of the shift is not theoretical.

According to McKinsey’s estimates:

By 2030, up to $1 trillion in U.S. B2C retail could be influenced or orchestrated by AI agents
Globally, $3–5 trillion in goods commerce could flow through agent-mediated decisions
These figures exclude services and B2B, suggesting further upside

At the same time, behavior is already changing:

Nearly 50% of users who try AI search prefer it over traditional search

Discovery is moving upstream — and humans are exiting the loop earlier than expected.

How Agentic Commerce Actually Works

Behind the scenes, agentic commerce relies on a new stack of protocols and capabilities that allow autonomy at scale.

1. Persistent Context (Model Context Protocol – MCP)

Agents need memory — not just prompts.

MCP enables:

Shared context across tools
Persistent intent over time
Reasoning that spans sessions, not single interactions

This is what allows an agent to act like a delegate, not a chatbot.

2. Agent-to-Agent Coordination (A2A)

Commerce increasingly becomes software negotiating with software.

A2A protocols allow agents to:

Discover each other’s capabilities
Coordinate tasks
Negotiate terms and execute jointly

Markets start behaving like distributed systems.

3. Agentic Payments (AP2)

Payments were built assuming a human decision-maker.

Agentic commerce requires:

Delegated authority
Verifiable intent
Programmable constraints (limits, rules, categories)

This introduces new trust constructs such as Know Your Agent (KYA) and machine-level mandates.

4. Computer-Use Agents

Where APIs don’t exist, agents operate interfaces directly.

This allows:

Interaction with legacy systems
Immediate reach without full replatforming
Faster adoption across fragmented ecosystems

What Changes for Businesses

1. Agents Become the Primary Customer

Soon, your “customer” is no longer a person browsing a site —
it’s an agent evaluating options on their behalf.

This means:

Brands must be discoverable to agents, not just humans
Differentiation must be legible to machines
Value propositions must survive algorithmic comparison

2. Funnels Give Way to Micro-Decisions

Traditional metrics — impressions, clicks, conversion rates — lose relevance.

Instead:

Decisions are continuous
Loyalty is delegated
Optimization happens at the intent level, not the page level

Commerce becomes probabilistic, not linear.

3. New Strategic Questions Emerge

Leaders must now ask:

Do we build our own agents — or optimize for others’?
How do we influence outcomes when humans aren’t choosing?
Where does brand power live when agents decide?

Trust, Risk, and Governance: The Hard Part

Autonomous systems introduce new failure modes.

Traditional controls assume:

Human review
Explicit consent
Static identities

Agentic commerce breaks these assumptions.

Key challenges include:

Delegated authority without loss of accountability
Fraud detection in agent-to-agent environments
Ensuring transparency, reversibility, and fairness

The core governance question shifts from:

“Is the system safe today?”
to
“Can the ecosystem manage what these agents may become?”

Organizational Implications

Technology & Data

Organizations will need:

Modular, interoperable APIs
Shared data standards
Context-aware, memory-driven personalization

Static rules won’t scale in an agentic world.

CX, Marketing, and Analytics

Customer experience teams must:

Design for agent interaction
Rethink loyalty and post-purchase engagement
Measure agent visibility, agent conversion, agent trust

Marketing becomes less about persuasion — and more about machine credibility.

What Leaders Should Take Away

1. This Is Already Happening

Major platforms are actively enabling agentic workflows across search, payments, and commerce.

This is not speculative.

2. Early Design Choices Will Compound

Organizations that:

Optimize early for agent interaction
Build trust and authorization into their systems
Expose clean, structured signals to agents

will compound advantages over time.

Those who delay risk becoming commoditized endpoints in someone else’s agent ecosystem.

The Strategic Question That Matters Most

The real question is not:

“How do we use AI in commerce?”

It is:

“What happens to our business when we are no longer the interface?”

At Madbro, we believe agentic commerce is not about replacing humans —
it’s about returning time, attention, and decision-making capacity by redesigning systems around delegation, trust, and intelligence.

We help companies unlock growth through AI, SaaS, and Automation.

We’ve partnered with the world’s leading innovators to redefine industries and create lasting impact. Whether leveraging AI agents to revolutionize customer experiences or designing automation solutions that scale, we build technology that generates real returns.