“Your next customer may never read the carefully crafted copy you’ve created. They won’t visit the apps you’ve built. They won’t care how your brand story makes them feel — because your next customer may very well be an algorithm.”
Katja Forbes dismantled that premise in the opening seconds of her DMS2026 keynote.
Forbes has 30 years of experience in customer experience and marketing. She is currently writing a book on AI agents and machine customers. Her session wasn’t speculation about a distant future. Instead, it was a frank assessment of what’s already happening in the market right now.
What Is a Machine Customer?
A machine customer is any AI or autonomous system that makes and executes purchasing decisions — on behalf of a human, or entirely without one. Forbes outlined three distinct types.
Delegated Agents learn a human’s values and preferences. They then go out and buy on their behalf. In a live demo, Forbes’s AI agent Tyler dropped a skincare brand the presenter had always bought. The reason wasn’t price or quality. The brand’s sustainability data simply wasn’t encoded in a machine-readable format. No website visit. No review browsing. Gone.
Autonomous Procurement Agents operate in B2B environments. “Node 741,” demonstrated live, manages logistics across 36 drone networks. It holds over $10 million in quarterly purchasing authority. Meanwhile, Walmart has been running AI procurement since 2022. Today, it closes roughly 70% of contracts with more than 2,000 vendors without a human involved.
Things as Customers are IoT devices that become buyers themselves — cars, printers, and smart home systems. Some Mercedes-Benz models already book their own service appointments, pay for parking, and negotiate charging autonomously. When a tire goes flat, the car knows first. It books the service, reroutes the trip, and may even order dinner for the drive home.
The Numbers Behind the Shift
The scale of this shift is hard to ignore. Consider these figures:
- 4,700% — Year-on-year increase in AI traffic to retail websites (2025 vs. 2024)
- 25% — Share of purchases Gartner expects AI to handle by 2030
- $30 trillion — Projected agentic commerce market size by 2030 (Gartner)
- 120 million — Orders processed via AI agents during Alibaba’s Liia Festival
As Forbes put it: “E-commerce completely reshaped retail. Agentic commerce is twice as large and arriving twice as fast.”
Three Layers of Agentic Commerce
Forbes mapped the current market into three distinct layers. Moreover, she argued that most businesses are investing in the wrong one.
Layer 1 — Foundation: Discoverability and Machine Readability
GEO, AIO, structured data, Schema markup. Around 90% of current vendor activity sits here. Forbes acknowledged it as essential. However, she was clear about its ceiling: “If everyone is discoverable, perhaps we’re all becoming interchangeable.”
"If everyone is discoverable, perhaps we're all becoming interchangeable."
👉 For a deeper dive into GEO and machine readability, see our recap of [How Marketers Are Using AI to Think Better, Not Just Work Faster →]
Layer 2 — Operational Trust
This layer covers Know Your Agent (KYA), verified transactions, and fraud prevention. Consumers are anxious about AI shopping — unauthorized purchases, identity issues, incorrect orders. As a result, building trust at this layer is necessary to stay in the game. That said, it doesn’t win the game on its own.
Layer 3 — The Stratosphere: Value Signals ★
This is where AI agents actually make their final decisions. Furthermore, it’s where Forbes believes the real competitive advantage lives.
Agents don’t read marketing copy. Instead, they query what is independently and verifiably true. Research across 250 brands and 40,000 AI queries found that 53% of the factors that determine whether an AI agent chooses your brand are outside your direct control — existing reviews, third-party certifications, and external databases.
There’s also a critical difference from human decision-making: AI agents are binary. A human can hold tension between “cheapest” and “most ethical.” An agent, however, cannot. It executes on the values it has been given. Meet the criteria, get chosen. Fall short, get dropped — with no second chances.
Two Scenarios That Show What’s at Stake
Forbes ran two live demonstrations. Both cut to the core of how machine customers behave differently from human ones.
Tyler drops a brand. The brand the presenter had always purchased failed a single test: its sustainability credentials weren’t machine-readable. There was no emotional history taken into account. No loyalty. No consideration. Simply eliminated.
A false certification costs everything. An organic cleanser brand had lost its certification 14 months prior. Nevertheless, it was still advertising “certified organic.” Tyler flagged the discrepancy by cross-referencing the official organic database. The brand was blacklisted immediately. Within 6 minutes, moreover, 312 other agents received the same data. Estimated impact: 68% drop in machine customer traffic, $1.47 million in lost agent-driven sales. As Tyler stated: “Integrity is binary. It’s either verified or it isn’t.”
The Funnel Is Collapsing
Forbes was direct: “Everything we have built at the human moment is going to fail at the machine moment.”
The reason is structural. AI agents move straight from intent to goal. In other words, the traditional funnel — awareness, consideration, evaluation, purchase — compresses or disappears entirely. There’s no lingering on a product page, no comparison browsing. If your brand meets an agent’s criteria before the query is made, you get chosen. If it doesn’t, you don’t even enter consideration.
Prompt injection — manipulating agents into selecting your brand through technical tricks — is not the answer either. “Tricks won’t get you there,” Forbes said. “We must actually evolve machine customer experience as a core part of what we offer.”
Questions the Industry Hasn’t Answered Yet
Forbes closed with open questions that discoverability checklists alone cannot resolve:
- Who is the real customer — Node 741, Tyler, the human behind Tyler, or the car?
- What does loyalty look like when an agent controls repeat purchases?
- How do you market simultaneously to humans and machines?
- How do you encode your organization’s values so that machines can read and verify them?
What to Do Starting Monday
First, encode your values as data, not copy. ESG claims, fair trade certifications, carbon footprint figures — these need to exist as structured, API-accessible, third-party-verifiable data. If an agent can’t query it independently, it simply doesn’t count.
Second, audit your agentic commerce readiness. Forbes shared a live “Value Signal” tool that takes your website’s ESG content, cross-references it against independent sources, and shows what’s verifiable — along with specific recommendations for improvement.
Third, build a cross-functional team. Agentic commerce is not a marketing problem alone. Therefore, data, IT, legal, and procurement teams all need to be part of this conversation from the start.
Final Thought
“What matters tomorrow is designed today. And you are the people designing how marketing is going to work.”
Agentic commerce will reshape markets the way e-commerce did — but bigger and faster. In that world, the standard for brand selection is no longer emotional resonance. It is verifiable truth.
This post is a summary and editorial reconstruction of the Katja Forbes keynote at DMS2026 (Digital Marketing Summit Seoul, March 24–25, 2026).