A B2A strategy is the roadmap that makes your business discoverable, understandable, and contactable for AI agents as potential customers. B2A – Business to Agent – describes a third source of demand alongside B2B and B2C: software that independently searches, compares, filters, and prepares purchasing decisions before a human even visits your website. Small and medium-sized enterprises (SMEs) that optimize exclusively for humans today are losing leads without even realizing it.
The data is clear. According to Adobe Analytics , traffic originating from generative AI services like ChatGPT or Perplexity on US retail websites increased roughly sevenfold between July 2024 and February 2025 compared to the same period the previous year. On the other hand, OpenAI reported approximately 800 million weekly active ChatGPT users in October 2025 – up from 700 million in August 2025. Both trends point in the same direction: the gateway to the web is shifting from search engines to assistants.
In my project work with South Tyrolean SMEs, I almost always encounter the same problem. The website is well-designed, appeals to people, and answers human questions. But as soon as an AI agent reads the same page, it lacks precisely the information it could provide: binding services, prices, languages spoken, accessibility, and responsibilities. The agent then recommends the company that provides this information – even if its design is inferior.
B2A Strategy: Why AI Agents Are a Source of Demand in Their Own Right
For years, we've thought in two directions: B2B, when a business sells to another company, and B2C, when a business sells to end consumers. B2A expands this thinking to include a customer that is neither a person nor an organization, but rather software acting on behalf of a person. This is precisely why it's not enough to simply treat B2A as a new channel. Business-to-agent requires its own dedicated business-to-agent strategy because agents select based on entirely different criteria.
- People decide about emotions: Trust, visual language, references, feeling. An agent doesn't evaluate those things.
- Agents decide on unambiguity: Services, location, language, availability, price range, and occupancy. If any of these details are missing, the business is eliminated from the selection.
- People read, agents extract: An agent doesn't take a statement out of context – they quote it, weigh it, and compare it with other sources about your business.
How quickly this shift will occur is uncertain. Gartner predicts in a 2025 forecast that by 2028, around 90 percent of B2B buying processes will be mediated by AI agents. This is a scenario, not a certainty – but nevertheless a clear indication of where the buying process is headed. Research, vendor comparison, and pre-selection are increasingly happening before a human even speaks to you.
B2A doesn't mean agents buy for you. B2A means agents choose – and your business needs to be among those choices.
AI agents as customers: What they need from your website
Agents are undemanding readers, but unforgiving scrutinizers. They don't need storytelling, animation, or visual language. They need facts that can be extracted, compared, and shared by machines. Those who take agents seriously as customers don't change the design, but rather the data quality of their website.
- A reliable factual basis: Services, locations, languages, opening hours, contact persons and prices in a binding format. Not hidden in block text, but clearly named and maintained.
- Machine-readable processing: Structured content, clear markup, and a separate, simplified version of the page. Our page explains the technical details. Agent Readiness.
- A genuine service catalog: Not a list of marketing phrases, but a versioned overview of available services with terms and conditions. The structure is described in the article on... Service catalog for AI agents described.
- An orientation file: credentials for llms.txt It provides language models with an organized overview of which content is relevant and where it is located. It's voluntary – but it costs little and creates clarity.
- Approvals: An internal decision determines which facts may be quoted. Prices and availability information without approval are useless to agents.
- Consistency: The same information appears on websites, in business directories, and in map services. Inconsistencies lead to an agent not mentioning the business at all, instead of choosing one of two options.
The 90-day plan: B2A for SMEs in five steps
A B2A strategy for SMEs rarely fails due to a lack of knowledge, but rather due to the sequence of steps. Starting with technology produces data clutter. Starting with a factual foundation builds a basis in 90 days that also works for GEO (Generative Engine Optimization) and traditional SEO.
Step 1 (Days 1–15): Clarify the factual basis
Write down everything an agent needs to know about your business: legal structure, locations, languages spoken, services offered, contact persons, availability, and pricing. Clarify internally who will authorize the confirmation of this information. Without this approval, any further steps are a shot in the dark.
Step 2 (Days 16–35): Making services machine-readable
Transfer the services into a structured service catalog. Each service receives a clear name, a brief description, target group, and conditions. Avoid parallel structures: one entry, one service provider, multiple distribution channels.
Step 3 (Days 36–55): Structuring prices and availability
You don't need to publish fixed prices. A price range, a "starting from" price, or a clear statement like "offer upon registration" is already valuable for agents. Add availability, lead times, and opening hours using the same format.
Step 4 (Days 56–70): Define approvals
Define which facts may be quoted and which may not. Regulate who implements changes and how versions remain traceable. This internal governance is the part that is most frequently lacking in practice.
Step 5 (Days 71–90): Set up the measurement device
Before launching any initiative, define how you will measure success: agent access, quote rate for fixed test questions, and queries originating from agents. Without these metrics, B2A remains an assertion rather than a controllable variable.
btlabs Core: the foundation for B2A
Everything described so far can be started with existing tools – a clean data model and an editorial process. However, if you're building a new website or fundamentally redesigning one, it's worth taking a look at the foundation itself. Our page on AI-ready websites with btlabs Core describes what this means in concrete terms.
btlabs Core delivers exactly the components you would otherwise have to painstakingly retrofit:
- llms.txt and brand.json as fixed files containing brand and structure information
- Markdown delivery the same content under stable URLs
- Single source (SSOT) For website, locations and future channels – maintain once, correctly everywhere
- Trilingualism at its core (German, Italian, English) with translated URLs and clean speech signals
- Hosting in the EU without cookie banners and without US trackers
The crucial point isn't the product, but the principle: Your facts exist in one place and are presented in multiple formats – for people, for search engines, and for AI agents. That's precisely the technical side of a B2A strategy. Our product btlabs demonstrates the detailed foundation.
Practical example from South Tyrol: before and after
A service provider from the Bolzano area, specializing in crafts and services, with seven employees, approached us with a question: Why are we recommended when someone uses Google, but not when someone asks an AI? The initial assessment was typical.
- Before: When asked in a test about a suitable business in the region, ChatGPT gave different names or answered evasively.
- Before: The responses contained outdated opening hours and a range of services that was two years old.
- Before: Prices and inquiry methods existed only as continuous text without a clear structure.
- Afterwards, after about 90 days: The test questions were answered correctly, with company name, services, and contact information from a reputable source.
- Later: Initial inquiries indicated that the company had been found via an AI assistant.
This isn't a lab result, but an observed trend from the project work. The crucial point isn't the number itself, but the direction: Before, the business was invisible to agents; afterward, it's quotable. The difference didn't arise from advertising, but from structure.
Measurement: How B2A becomes a controllable quantity
B2A is only a strategy if you can measure it. Four metrics are enough to start with, and none of them require complex tracking.
- Agent accesses in the server log: Requests from common bot and agent identifiers such as GPTBot, PerplexityBot or ClaudeBot, counted per month.
- Quotation rate in fixed test questions: Formulate 10 to 15 questions that a customer would actually ask, and check monthly if and how your business is mentioned. How we document this is explained below. LLM monitoring for AI visibility.
- Inquiries originating from agents: A simple field in the contact form – “How did you find us?” – is enough to identify patterns.
- Factual consistency: Are opening hours, services, and prices consistent across all channels? Every discrepancy costs recommendations.
The same systematic approach is crucial every month. Individual samples are meaningless; repeatable test runs reveal trends.
What B2A 2026 is not yet
I believe it's only fair to clearly define the development stages. AI agents that can independently book appointments or place orders are technically ready, but not yet standard. The same applies to live price calculators, appointment displays, secure customer areas, or a shop and inventory module built on the same foundation. Anyone who talks about fully automated agents as the norm today is selling a promise, not a solution.
The reliable part today is unspectacular yet effective: machine-readable facts , a well-maintained service catalog, clear approvals, and regular measurement. This ensures you're prepared when agents actually start acting – without having to invest money in technology that no one needs yet.
What this means for you
B2A doesn't change your business, but rather the way demand is generated. Your brand remains important—to people. For agents, what matters is that your facts are clear, complete, and citable. A side benefit: A solid factual foundation simultaneously improves your AI visibility, your geolocation, and the quality of your website for human visitors.
If you don't know where you stand, the most honest starting point is to assess your current situation: What facts exist about your business today, in what form, and who confirms them? Everything else builds upon that.
Questions and answers about B2A
What exactly is B2A – Business to Agent?
B2A refers to business relationships where AI agents act as the buyer: they search, compare, filter, and prepare transactions. A B2A strategy makes your business discoverable, unique, and requestable by these systems. Thus, B2A complements the traditional B2B and B2C channels, but does not replace them.
Do I need a completely new website for B2A?
No. In many cases, it's sufficient to clarify the factual basis and deliver the most important content in a machine-readable format. A new website only makes sense when structure, multilingualism, and maintenance effort are no longer viable – then a foundation like btlabs Core is a cleaner choice for SMEs than another rebuild.
How do I measure whether AI agents are using my website?
The easiest way is via the server log: There you can see requests from bots and agents like GPTBot or PerplexityBot, counted per month. Additionally, a monthly test run with fixed questions to several AI assistants is helpful. Together, these two metrics show whether agents can reach you at all and whether they are correctly representing you.
Is GEO the same as B2A?
No, even though they are related. GEO ensures that generative systems correctly identify and quote you in responses. B2A goes a step further and asks whether your business is requestable and selectable for a trading agent – in other words, whether services, prices, and availability are structured and readily available. GEO is therefore the visibility level, B2A the supply level.
What does it cost to get started with a B2A strategy?
The first step mainly takes time, not BudgetA clear factual basis and a well-maintained service catalog can be built in-house. Investments only arise during the technical delivery phase, for example through Markdown versions. llms.txt or a new foundation. It makes sense to start with a consultation that realistically assesses the sequence and effort involved.
Are AI agents already able to book for me today?
While some systems are technically capable of this, it's not yet standard practice. The tangible benefit today lies in agents accurately understanding and recommending your business. Everything beyond that – booking agents, price calculators, appointment display – are development stages built upon a solid foundation of facts.
Sources
- Adobe Analytics — business.adobe.com (2025)
- Gartner — gartner.com (2025)
- OpenAI — openai.com (2025)
