Imagine planning a beach weekend with friends without anyone actually doing the planning. A software system quietly tracks surf forecasts for Bali, and the moment conditions look ideal, it searches flights, checks accommodation, and packages everything into a single proposal for the group. Once everyone taps “approve,” the same system finishes the booking on its own. No one compared ten browser tabs of hotel listings. No one argued with an airline chatbot at midnight.
That scenario isn’t speculative marketing copy — it’s the real opening example used by Tuomas Peltoniemi, Managing Director and Design & Digital Products Lead at Accenture Song Southeast Asia (the customer-experience and marketing arm of the global consulting firm Accenture), during a recent industry session called “Owning the Experience: CX for the Agentic Era.” It illustrates a shift that matters enormously for anyone building income online: the customer placing the order is increasingly not a person at all.
For freelancers, solo e-commerce sellers, digital nomads, and small service businesses, this creates a genuinely new income opportunity — and a genuinely new risk. If an AI agent (a software assistant that can research, decide, and act on someone’s behalf — distinct from a basic chatbot that only answers questions) is now the one choosing which freelancer gets hired or which product gets bought, then being “readable” and “chooseable” by that software is becoming as important as impressing a human client. Here is what the underlying research found, and what it actually means for how solo earners should be positioning themselves.
Customers Are Quietly Outsourcing Their Judgment
The session drew on Accenture’s Consumer Pulse Research, a survey of 25,590 consumers across 16 countries conducted in January 2026. The headline number is startling even to the researchers who found it: 74% of respondents said they trust their personal AI agent more than their closest friend when it comes to making a purchase decision on their behalf. On the business side, 70% of executives expect AI agents to take over marketing and sales decisions directly within the next three years.
This is not a small behavioral tweak. For twenty to thirty years, commerce has repeatedly redefined who “owns” the customer relationship. In the industrial era, companies decided what to sell and how; customers chose from what was in front of them. When the web arrived, buyers became their own researchers, opening dozens of tabs to compare prices themselves. Today, many people are simply tired of that labor — a growing number have started replacing routine Google searches with a large language model (LLM), the technology behind tools like ChatGPT or Claude, asking it to recommend products directly. The AI agent is stepping into the middle of that relationship as the new curator of the customer experience.
Not Every Purchase Gets Delegated the Same Way
The shift isn’t uniform, and that distinction matters if you’re trying to earn money in this new environment. Accenture’s data shows 85% of consumers are comfortable co-deciding a purchase with AI, and 74% are comfortable letting an agent act on their behalf for tasks like negotiating a deal or resolving a complaint. Only 9% are ready to hand over both the decision and the payment to an agent fully autonomously today — but that number is expected to climb.
To map where different categories sit, Accenture built what it calls the Delegation Dial — a way of visualizing how willing consumers are to hand control to AI for different kinds of spending. At one end sit emotional, high-stakes purchases: skincare, beauty products, travel. Buyers want AI to help them think, but not to decide for them. At the other end are recurring, low-drama purchases like subscriptions and weekly household restocking, where many people have already let an agent run the whole thing.
One Hong Kong-based millennial surveyed for the research put the line clearly: letting an AI agent book the flight is fine, but choosing the hotel room is not, because small details like the view or the layout still matter to her personally. Even a single purchase journey can straddle both ends of that dial. In the example of a first-time home buyer, choosing a neighborhood remains too emotional to delegate, and submitting an actual offer is something people still insist on doing themselves — but calculating affordability or arranging mortgage insurance is exactly the kind of task many are now happy to give to an AI agent.
For a freelancer or solo seller, this is a practical filter: services and products that are emotionally loaded and identity-driven still need to win over a human. Services that are routine, comparative, and price-driven are increasingly won or lost based on whether an AI agent can even evaluate the offer.
Stop Building for Channels — Build for the Whole Journey
Peltoniemi’s third point is organizational, but it applies just as much to a one-person business as to a bank. Many companies still design their websites and apps as if each channel were a separate silo — one team for the mobile app, a different team for the website — when what actually needs to be “owned” is the customer’s entire journey, regardless of how they arrive.
Accenture Song frames the fix as three shifts: from thinking in channels to being present wherever a customer’s AI agent chooses to make contact; from building service one channel at a time to what it calls Orchestrated Intelligence — a single layer that listens, decides, and connects a customer to whatever they need; and from rigid, pre-set customer paths to Composable Capabilities, meaning a business builds a function once and lets every channel or partner call on it.
A bank illustrated the payoff on stage: if a back-end agent can see a customer’s salary cycle, credit score, and existing products, it can proactively message the household through chat that they are financially ready for a first home — before anyone applies. The bank gets the lead without waiting; the customer gets an answer without filling out a form. Take it one step further and the same bank could open its services so outside agents can plug in too — meaning a customer’s entire moving-day checklist, from utilities to internet to a moving company, gets handled through one conversation with their personal AI agent, instead of five separate phone calls. Marriott was cited as an early example of a brand already letting personal AI agents book rooms according to a traveler’s known preferences.
The practical lesson for a solo operator: your Fastwork profile (a popular Thai freelance marketplace, similar to Upwork or Fiverr), your Instagram shop, your LINE Official Account — known in Southeast Asia as LINE OA, a business messaging tool built on the LINE chat app that functions much like WhatsApp Business — and your own website are not separate storefronts anymore. An AI agent evaluating you doesn’t care which one it lands on; it cares whether the information is consistent and complete everywhere.
The New Job Description: Getting Read by a Machine
This is the section that matters most for anyone monetizing with AI, because it names an entirely new, sellable skill. Peltoniemi calls it computational legibility — whether an AI agent can actually read, parse, and understand what a brand or seller offers. When Oreo, a cookie brand nearly everyone recognizes, tested its own website against this standard, only 10% of AI agents could read it correctly. If a company with that level of global recognition struggles, most small sellers almost certainly have the same blind spot.
The consequence of being unreadable isn’t a lower ranking — it’s disappearing from the shortlist entirely. Sellers without structured product data (information formatted in a standardized way machines can parse, rather than buried in a photo or a paragraph of marketing copy) and businesses that leave certifications and guarantees as unverifiable design elements on a webpage simply don’t show up when someone asks ChatGPT or Claude for a recommendation. Two other factors carry heavy weight: pricing transparency and a verifiable delivery track record. AI agents are already cross-checking claims against real customer feedback — OpenAI’s direct integration with Reddit was cited as an example of a system that can check, in milliseconds, whether what customers say about a brand matches what that brand claims about itself.
Two things are immediately actionable, and both are things a freelancer could learn to sell as a service to other small businesses: running an AI Discoverability Audit — checking whether AI crawlers are even allowed to read a given website, since many brands are unknowingly blocking them — and converting prices, fees, and eligibility terms into machine-readable product schema (structured metadata search engines and AI agents can parse directly). This is essentially the AI-era evolution of search engine optimization, sometimes called generative engine optimization (GEO) or answer engine optimization (AEO) — and demand for people who understand it is only going to grow.
แนะนำ
None of this replaces the human side. A brand optimized purely for machines but meaningless to people still loses whenever a human’s judgment is the final filter — and a brand that wins hearts but stays invisible to agents never even makes it onto the list to be judged in the first place. Both jobs have to be done at once.
Loyalty Now Comes With Conditions
The last research finding should worry anyone who has built a business on repeat customers. Peltoniemi used his own two decades of buying Nike sneakers as an example — yet admitted that if an AI agent told him another brand suited him better today, there’s a real chance he’d believe it. The data backs this up: 37% of behaviorally loyal customers — people who have always chosen the same brand — said they would still switch if an AI agent recommended something else that better matched their needs. A promise a brand fails to deliver on is exactly the kind of gap that sends an agent looking elsewhere.
For solo sellers, this cuts against the old assumption that a loyal client base is a safe moat. It isn’t, anymore — not unless your information stays consistent everywhere you have a presence, and you keep showing up specifically when you’re genuinely the right fit, rather than everywhere all the time.
What This Looks Like Across Southeast Asia
The same dynamics are unfolding, at different speeds, across the region. In Vietnam, sellers on e-commerce platforms like Tiki and Shopee are already competing on structured listings and verified shipping performance in ways that map directly onto “computational legibility.” In Indonesia, GoTo’s Tokopedia marketplace and messaging-based commerce through apps like WhatsApp face the same channel-fragmentation problem Accenture Song describes — a seller’s WhatsApp catalog, Instagram shop, and marketplace listing often tell three slightly different stories about price and stock.
In Thailand specifically, a freelancer or small seller typically has a presence split across Shopee and Lazada (the two dominant Southeast Asian e-commerce marketplaces), a LINE OA for direct customer chat, and often a listing on Fastwork if they sell services. Payments frequently run through PromptPay, Thailand’s national QR-code-based instant bank transfer system. None of these platforms currently talk to each other, and almost none of them expose data in a way an AI shopping agent can cleanly read. That gap is exactly where an early-mover income opportunity sits: helping small Thai and Southeast Asian businesses become legible to the AI agents their own customers are starting to rely on.
The Actionable Takeaway
Whether you’re in Bangkok, Ho Chi Minh City, Jakarta, or anywhere else building income around AI tools, the opportunity isn’t just “use AI to write content faster.” It’s understanding that AI agents are becoming a second customer you have to satisfy — one that reads structured data, checks reviews for consistency, and has no loyalty to you unless the facts stay straight. Start by auditing whether your own listings, profiles, and pricing are consistent across every platform you sell on. Then consider that the same audit is a service millions of small businesses in this region will soon need — and very few people currently know how to sell it to them.
Key Takeaways
- AI agents, not just human buyers, are increasingly the ones evaluating and choosing sellers, freelancers, and products.
- 74% of consumers say they trust a personal AI agent more than their best friend when it comes to purchase decisions.
- Willingness to delegate varies by category: emotional purchases stay human-led, routine ones are handed to AI.
- “Computational legibility” — whether AI can actually read your listings and pricing — is a new, sellable freelance skill in Southeast Asia.
- Even loyal customers will switch brands if an AI agent recommends an alternative, so consistency across platforms now matters more than ever.
Frequently Asked Questions
Q: Can AI agents really decide what I buy without me choosing?
A: Not entirely yet, but research shows most consumers already let AI co-decide low-emotion, routine purchases, and a small but growing share let agents handle both the decision and payment automatically.
Q: What exactly is an “AI agent” as opposed to a chatbot?
A: A chatbot mostly answers questions, while an AI agent can independently research options, make a decision within set boundaries, and complete an action like booking or paying on a person’s behalf.
Q: How can a freelancer or small seller in Thailand make money from this trend?
A: By learning to make listings and pricing “machine-readable” and offering that as a paid audit or fix-up service to other local businesses that are currently invisible to AI shopping agents.
Q: What is an AI Discoverability Audit?
A: It’s a check of whether AI crawlers and agents are even able to access and understand a website or online listing, which many businesses unknowingly block or leave unreadable.
Q: Why did an AI agent fail to understand a brand as recognizable as Oreo?
A: Because its product information wasn’t formatted in a structured, machine-readable way, even though it was perfectly clear to human visitors.
Q: Does this mean SEO is dead?
A: No, but it is being supplemented by a related discipline sometimes called generative engine optimization or answer engine optimization, focused on being understood by AI systems rather than just ranked by search engines.
Q: Will loyal customers still choose my business over competitors?
A: Not automatically — the research found over a third of habitually loyal customers would switch if an AI agent recommended a better-matching alternative.
Q: What is the “Delegation Dial” mentioned in the research?
A: It’s a concept describing how comfortable people are letting AI handle a purchase, ranging from full human control for emotional buys to full automation for repetitive, low-stakes ones.
Q: How does LINE OA relate to AI agents choosing sellers?
A: A LINE Official Account often holds different pricing or stock information than a seller’s marketplace listing, and inconsistencies like that can cause an AI agent to distrust or skip the seller entirely.
Q: Is this shift only happening in large companies, or does it affect small solo sellers too?
A: It affects solo sellers directly, since AI shopping agents increasingly filter and recommend based on structured, verifiable listing data regardless of business size.
Q: What’s the fastest first step to prepare for this?
A: Check that your pricing, product details, and reviews are identical across every platform you use, since inconsistency is one of the main reasons AI agents skip a seller.
Q: Is this trend specific to Thailand, or is it happening elsewhere in Southeast Asia too?
A: It’s regional — similar patterns are already visible on e-commerce platforms in Vietnam and Indonesia, though adoption speed varies by market.
