Key Takeaway
A new form of competition is emerging for Georgian businesses. If consumers increasingly delegate product and service choices to ChatGPT, AI search systems or shopping agents, companies will no longer compete only on advertising, website design or social-media reach. A new question becomes critical: can an AI system find the company, understand its offer correctly, verify the information and compare it with alternatives?
This shift is already becoming practical. In March 2026, OpenAI expanded product discovery in ChatGPT and connected it to structured merchant catalogues through the Agentic Commerce Protocol. OpenAI says shopping experiences can compare products using price, features, reviews and other information, while merchant selection can consider factors such as availability, price, quality and whether the merchant is the maker or primary seller. Direct product feeds can provide up-to-date titles, descriptions, images, prices and availability. Shopify merchant data is also integrated through Shopify Catalog.
The implication is broader than one platform. Google Search has long encouraged product structured data that explicitly describes price, availability, shipping and returns. The digital market is moving toward a world in which visibility increasingly depends on whether the information about a business is not only persuasive to a human but also unambiguous to a machine.
Competition begins before the answer
In the traditional web, consumers often performed the comparison themselves. They opened a search engine, clicked several links, visited websites, read product descriptions and eventually made a choice.
AI-mediated search can compress that journey.
A consumer may ask one question: Which Georgian wine should I buy as a gift under GEL 80? Which hotel in Kakheti is best for a family? Which local cosmetics brand offers fragrance-free products? Where can I buy a specific device in Tbilisi today?
The AI system can search, compare and narrow the options before the user sees them.
The consumer may never visit ten competing websites.
That creates a new risk: a company can exist online and still fail to appear in the answer.
That does not imply deliberate exclusion. The reason can be operationally simple: unclear product descriptions, prices embedded only in images, stale inventory, missing delivery times, hidden return policies, inconsistent information across channels or unstable product naming.
Humans can sometimes resolve those contradictions.
Machines need clarity.
| What the AI needs to understand | Ambiguous version | More useful version |
| Product | “Premium Georgian product” | Stable product name, category and factual attributes |
| Price | Price visible only in an image | Final price in readable page/feed fields |
| Availability | “Contact us for availability” | In stock / out of stock / preorder |
| Delivery | “Fast delivery” | Defined area, timing and cost |
| Returns and warranty | Hidden or vague terms | Public, clear policy with a durable link |
| Business identity | Brand name only on social media | Official site, contact/legal information and consistent identity |
Brand language is no longer enough – machines need factual language
Traditional advertising often relies on evaluative phrases: best, premium, unique, high quality, most comfortable.
Those claims may influence human perception. They are difficult for a machine to compare.
For an AI system, factual attributes are more useful.
“Best hotel in Kakheti” is weak information. Location, room type, nightly price, breakfast, parking, cancellation policy, child facilities and distance from attractions are comparable.
“Highest-quality Georgian wine” is less machine-readable than grape variety, vintage, region, alcohol content, bottle size, price, food pairing and verified origin.
“Trusted online store” is a self-description. Public legal identity, contact information, full price, delivery time, return policy, warranty and verified reviews are stronger signals.
This does not make branding obsolete. Emotional brand value still matters to people. But AI-mediated commerce requires a factual layer underneath the brand – one that machines can parse and verify.
Georgia’s consumer-protection framework reinforces the same direction. The Georgian Competition and Consumer Agency states that remote customers must receive clear information on terms, costs and withdrawal rights. Except for legal exemptions, consumers generally have a 14-day withdrawal right for remote purchases. The agency’s 2026 decisions continue to enforce the obligation to provide adequate consumer information.
In other words, much of what makes a business machine-readable is also good consumer practice.
What “machine-readable” means in practice
Being understandable to a machine does not mean writing a website for robots and forgetting people.
The best commercial information is clear to both.
A product page should have a stable official name, concise factual description, price, availability, core attributes, images, delivery terms, return policy and a working URL. Data across public channels should be consistent.
This principle is becoming technically standardized. OpenAI’s Agentic Commerce Protocol supports structured product feeds with explicit product fields. Google Search Central separately structures price, availability, shipping and return information through Product and Offer markup.
The key lesson for a Georgian company is not to blindly adopt one technical specification. It is to build a reliable data architecture: one authoritative source of product and service information from which every public channel can stay synchronized.
If a website says an item is in stock, social media says it is sold out and a comparison service shows an old price, the information environment is low quality. An AI system may simply prefer a competitor whose offer is easier to verify.
| Information layer | Why it matters | What the business should do |
| Official product page | Creates an authoritative source | Use one stable URL, name and factual description |
| Structured data | Makes key fields explicit | Mark price, availability, offer, shipping and returns where supported |
| Product feed | Supports catalogue updates | Use the relevant platform feed when available |
| Reviews | Adds external experience signals | Show authentic reviews and responses |
| Policies | Reduces uncertainty | Publish clear return, warranty and delivery terms |
| Freshness | Stale data damages comparability | Synchronize price, stock and terms across channels |
This can be an opportunity for small Georgian businesses
AI-mediated choice does not automatically favor large companies.
In some situations it can reduce the brand-awareness advantage.
A human may know only the largest brands. An AI system may surface a smaller offer if the product fits the user’s constraints more precisely and its information is complete.
That creates an opportunity for niche Georgian businesses: small hotels, family wineries, designers, cosmetics makers, specialist retailers, food producers and professional service firms.
When the consumer asks a highly specific question, broad brand recognition may be less important than exact fit.
“Most famous guesthouse” and “guesthouse in Sighnaghi with a four-person room, free parking, breakfast and late check-in” are different queries. The second rewards structured, specific information.
In the AI era, a small brand may not need to become the loudest option for everyone.
It may need to become the clearest option for the right request.
Trust cannot be created only in website code
There is a risk that businesses interpret this shift as a purely technical task.
It is not.
If a product’s real price changes but the feed is stale, structured data distributes the error more efficiently. If the return policy is not honored, describing it perfectly does not create trust. If reviews are unreliable, AI optimization cannot substitute for credibility.
OpenAI’s own commerce guidance emphasizes factual product copy, durable public links, correct variant modeling, and up-to-date price and availability information.
That is an important signal: the new visibility layer depends on data quality, not simply content volume.
Ultimately, AI systems also need trustworthy sources. Inventory management, delivery, service, returns and warranties therefore become part of discoverability.
Agentic commerce raises the standard further
If AI only writes an answer, ambiguous information has limited consequences: the user can still verify details manually.
If a shopping agent begins to act – selecting a product, preparing a cart or initiating a purchase – the need for accurate information becomes much stronger.
The system needs to know not only what the product is, but whether it is actually available, where it ships, the final price, delivery timing, return rights and which specific variant is purchasable.
OpenAI’s 2026 commerce infrastructure points in this direction. The Agentic Commerce Protocol is designed as a connective layer between merchants and ChatGPT users, using structured catalog data so products can be understood and surfaced in context.
According to an assessment by BTU researchers, this could shift one of the main units of digital competition. Businesses first competed for page views, then for search ranking. In agentic commerce, competition may increasingly be about whether the business enters the AI system’s shortlist at all.
This does not mean advertising or traditional search optimization disappears.
A new layer is being added: machine-readable trust.
What Georgian businesses should do now
The first step is not a new advertising campaign.
It is an information audit.
A business should compare how the same product or service is described on its own website, social platforms, search results, marketplaces and other public sources. Conflicting prices, outdated contact information, discontinued products or unclear returns are now both customer and AI problems.
Second, product and service descriptions should move toward factual language. What is it? Who is it for? What does it cost? Where is it? When is it available? What is included? What are the limitations? What are the return or warranty conditions?
Third, businesses should use structured data where their technical platform supports it. For online retail this means technically correct product, price, inventory, shipping and return information, and product feeds where relevant.
Fourth, the official website should become the authoritative public source for the brand. Critical facts should not exist only in temporary social posts or images.
Fifth, results should be measured. “AI visibility” should not become another vague marketing slogan. Businesses should track referral sources, recurring AI-originated questions where measurable, data accuracy and the conversion of such traffic or inquiries into sales.
| Step | First action | What to monitor |
| 1. Information audit | Compare website, social channels, marketplaces and search results | Conflicts in names, price, stock and terms |
| 2. Factual description | Add concrete attributes beneath marketing language | Whether pages answer the user’s core questions |
| 3. Structure the data | Expose key product and offer fields technically | Missing fields, errors and update speed |
| 4. Authoritative source | Build complete canonical pages on the official site | Whether critical facts exist only in posts or images |
| 5. Measure outcomes | Track AI/search referrals and inquiries where measurable | Leads, sales, data accuracy and repeat discovery |
Georgia’s advantage may be simplicity
Large companies often have complex catalogues, multiple legacy systems, conflicting data sources and slow approval processes.
A small Georgian business may have the opposite advantage: one catalogue, one responsible owner and the ability to correct information quickly.
A small winery may only need ten products described perfectly.
A boutique hotel may need rooms, prices, availability and policies to be consistently represented.
A professional-services firm may need a clear description of services, target customers, process, outcomes and contact path.
In the AI era, a small business may compete through clarity rather than scale.
The biggest risk: the brand exists, but the machine cannot understand it
Existence and visibility have never been the same thing online.
AI makes the gap more consequential.
A company can be genuinely good, but if its information is incomplete, contradictory or outdated, an AI agent may struggle to establish that quality.
A competitor may not have a clearly better product. It may simply be easier to understand, more current and easier to compare.
The new competitor may therefore be the company’s own information disorder.
Conclusion
For Georgian businesses, AI discoverability may soon become as practical an issue as search visibility, online payment or social-media presence.
The mistake would be to treat this merely as writing copy “for AI.”
The strongest foundations for appearing in an AI answer are the same foundations that create good business for humans: accurate product information, full pricing, real availability, clear delivery terms, understandable returns, credible reviews and consistent identity.
If consumers increasingly say, “Choose for me,” competition changes.
A brand will no longer need only to persuade the person.
It will need to maintain facts that allow the machine to conclude that this company is a relevant, verifiable and trustworthy option for that person’s request.
Data and Main Sources
OpenAI, Powering Product Discovery in ChatGPT, March 24, 2026.
OpenAI Help Center, Shopping with ChatGPT Search, updated 2026.
OpenAI Developers, Agentic Commerce Protocol, 2026.
Google Search Central, Merchant Listing Structured Data – Product and Offer.
Georgian Competition and Consumer Agency, current consumer-rights guidance and 2026 decisions.
BTU research and analytical materials on agentic commerce, digital business and the Georgian market.
Prepared by the academic team of Business and Technology University and the BTUAI Research Team, Tbilisi, Georgia.