By Himanshu Sinha | AI & Machine Learning Leader | Expert in AI-Driven Product Development, Personalization, and Business Strategy.
Sometime in the next few years, a hotel room will be booked with no search, no ad impression, no website visit and no phone call. A software agent will read pricing data, compute the cash value of loyalty benefits, weigh cancellation terms against its owner’s calendar risk and confirm the reservation before she finishes her coffee. The hotel’s marketing team will never know the sale almost went elsewhere—or why it didn’t.
I’ve spent the better part of a decade building personalization and lifetime-value models for large loyalty programs. Everything my field has built—propensity models, campaign optimization, remarketing, offer sequencing—rests on an assumption that’s so basic, we never state it: A human’s on the other end of the message. That assumption is starting to break, and most enterprises aren’t measuring the break.
The work moves from the customer to the machine.
Take a traveler I’ll call Sarah, a top-tier elite member who logs 11 business trips a year. Today, she does the work herself: searching, comparing, second-guessing, deciding whether her points are worth burning. Every step is a surface where a brand can persuade her.
Give Sarah a capable personal agent and one instruction—”San Diego, three nights, inside policy, beachfront if possible, under $2,000 all in”—and the journey collapses. The agent evaluates thousands of properties in seconds. It remembers the hotel that botched her late checkout twice. It computes what her elite status is actually worth at each candidate, cross-references her meeting locations and the weather and presents one recommendation. She approves it without ever seeing a search results page.
Every persuasion surface the industry spent 30 years perfecting sits on a journey the customer no longer takes.
This isn’t ‘another channel.’
SEO taught companies to be findable by machines that ranked pages for humans to read. The agent economy demands something harder: being evaluable by machines that decide on humans’ behalf. Google influenced which options you saw. An agent chooses which option you get.
That changes what wins. Lifestyle photography and emotionally resonant copy don’t move agents. They respond to structured, verifiable value such as transparent pricing, computable loyalty benefits and machine-readable cancellation terms—reliability that can be scored from data. Call these machine-readable brands.
In an agent-mediated comparison, a $290 property with a thin loyalty proposition can lose to a $320 property whose benefits and flexibility compute to higher expected trip value, and no brand tracker will ever explain why.
Loyalty gets more valuable, not less.
The conventional fear is that agents commoditize everything. I’d argue the opposite for one reason: Agents have perfect memory. A human member vaguely recalls her points are “worth something.” Her agent knows exactly what elite status returned last year in upgrades and breakfasts to the dollar and prices that into every decision. A brand that consistently delivers earns something new: agent loyalty. Unlike human loyalty, a rival’s clever campaign can’t erode it. Only better math or worse service can.
Measurement breaks, too. Customer lifetime value assumes a person making incremental choices, but one agent may soon route a household’s entire travel, dining and financial-product spend. We’ll need a companion metric, agent-controlled lifetime value (ACLV), capturing the economic activity a brand wins or loses through an agent’s standing evaluation.
What should hospitality incumbents do now?
If I ran strategy at a global hotel company, four moves would be on this year’s road map:
1. Audit machine perception. Ask today’s leading AI assistants about your prices, benefits and policies, and catalog what they get wrong. That error rate is tomorrow’s lost revenue.
2. Make loyalty computable. Publish benefit values, upgrade odds and redemption math in structured form an agent can verify because opaque “subject to availability” language reads as zero.
3. Instrument the funnel to identify AI-assisted sessions and treat them as a distinct, growing segment with its own conversion logic.
4. Prepare revenue management for machine-to-machine negotiation—an agent presenting constraints, a pricing engine countering with a dynamic package, a deal closing in milliseconds.
The pieces exist on both sides. What’s missing is the handshake, and someone will build it first.
There’s a governance question boards should raise before regulators do: If a few foundation models mediate a large share of consumer choice, ranking logic and paid placement inside them become market-structure questions. Companies that build trust signals agents can verify will fare better than those that try to game agents the way some once gamed search.
For a century, companies competed for consumers. For the next one, they’ll compete for the systems that represent consumers. The most valuable customer of the coming decade may never open your app or click on your ad. By the time it appears in your dashboards as “direct / no referrer,” the invisible economy will already be deciding who wins.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?