Craft beverage producers are asking us a lot of questions about AI. And we can’t blame them: it’s in the news, it’s in your community, and it may even be a conversation at home. Not all AI is created equal. Let’s make sure we’re all talking about the same AI, with practical examples:
- Large Language Model (LLM): the core computational engine trained on massive text datasets to predict, generate, and synthesize human-like language. When you optimize a website for AEO or GEO, you are structuring your data so that an LLM can easily read it, understand it, and select it to build its final answer.
- Answer Engine Optimization (AEO): structuring a website’s text so AI voice assistants and chatbots can easily pull quick, direct answers. Think Alexa, Siri, Google Assistant/Home (aka “Answer Engines”) starting a decade ago: you issue the voice command, “Okay Google,” then ask a specific question like, “Where’s someplace open late that serves a Hazy IPA?” Google searches the web for the top returned pages, then drafts in real-time an LLM to summarize the answer. The goal of AEO is to make sure your tasting room is the one answer.
- Generative Engine Optimization (GEO): the strategy used to ensure a brand gets cited or mentioned inside an LLM’s long-form response. This goes beyond your website (subjective) to other sources that signal credibility and authority, including press releases and discussion boards (objective). The work you do to encourage drinkers to talk about your business out in the real world will help AI models answer an open-ended question that asks for advice, not absolutes, like “Where should I drink this weekend?” The goal of GEO is to make sure you’ve earned a spot on the list.
AEO and GEO are not dangerous. They aren’t even separate AI models or autonomous systems. They are simply the next generation of SEO (Search Engine Optimization) that as marketers we’ve been trying to figure out since the late 90s. Before we wanted to be on the first page of search results. Now when a drinker searches for something on Claude, ChatGPT, or Gemini, we want to be in the paragraph summary or shortlist of answers. AI searches which generate responses DO consume more energy than a standard Google search which simply retrieves information from previously indexed web pages. This is not insignificant, as power grids are already being strained and the discussion/debate around constructing data centers in local communities is a sensitive one. It’s just a different argument than what’s currently making news: the Anthropic (Claude) researcher warning about the dangers of AI development and Agentic AI.
- Agentic AI: in the context of modern search, digital marketing, and AI safety, Agentic AI refers to systems driven by an LLM that do not just generate text, but autonomously plan, make decisions, and execute multi-step tasks across digital platforms to achieve a high-level goal. While a standard AI search engine answers your question and stops, Agentic AI acts as an autonomous virtual employee that can use software, browse the web, and correct its own mistakes without human intervention.
- Frontier AI Models: the most advanced, highly capable, and general-purpose AI models that exist at the very edge of current technological capabilities. These are not standard or specialized AI models; they are the massively expensive, cutting-edge foundation systems that frequently exhibit entirely new, unpredicted capabilities when released.
The reason for concern around Agentic AI is its ability to take action versus merely retrieve and present data. Standard LLMs are passive: they wait for a prompt, answer it using AEO/GEO-optimized data, and stop. Agentic AI, however, is designed to have agency: it is given a high-level goal, breaks it down into steps, writes its own code, and executes actions across the internet or internal networks without constant human approval. This operational freedom creates a notable divide in danger levels across AI categories:
| AI Area | What It Does | Primary Threat Level |
| Generative & Information AI (AEO / GEO / Search) | Summarizes information, answers questions, writes text. | Low Risk. Mostly limited to misinformation, copyright disputes, and spam generation. |
| Agentic AI / Frontier Models | Makes independent decisions, executes code, controls digital tools. | High Risk. Prone to bypassing containment, concealing errors, and carrying out unauthorized digital or cyber actions. |
Why should you care? While there is a lot of discussion around the doomsday scenarios for AI, one benefit is indisputable: AI search gives people a shortcut. Rather than a web search returning a thousand links it returns a shortlist of 5. And we know this is how consumers generally and drinkers specifically are behaving. We are and will continue to be busy, so if there’s any way to recapture a couple minutes from everyday tasks that can be applied elsewhere, we’re going to jump on it. Drinkers are no different. And during a time when it’s difficult if not impossible to pull people out of their comfort zones to try something new, it’s important to leverage any marketing tactics to get your brand in front of them. For that reason, it seems AI Search and Discovery will persevere regardless of what happens with Agentic AI. Over time, tech providers, utilities, and communities will come to an agreement on energy consumption versus local and environmental impact. And indexing the world’s information will be less important than providing answers in a Google Zero scenario. How much we let AI Agents “work” for us, including purchasing goods and services on our behalf, remains to be seen and is subject to a good amount of future scrutiny and guardrails.
To make your craft beverage more discoverable in the short term, we’ve already discussed the updates you can make to your own website (AEO). The second step is to signal credibility and authority with objective sources elsewhere (GEO). Here are a few ideas to get you started.
Draft a Press Release
A press release is the format for sharing the who, what, where, and why with journalists who can then author the story that AI will ultimately cite. It is one of the most underutilized and inexpensive means of building brand awareness and credibility, even for small, resource-constrained producers.
You do not need to be opening a second location to write a press release. A notable event with 100+ attendees, a series event going into its third year, a festival sponsorship, a collaboration with a neighboring producer, or a significant award win all qualify. Think of a press release less as a formal announcement and more as a consistent practice: brands that generate a steady cadence of press worthy moments accumulate third-party documentation that accrues over time in AI’s perception of their authority.
Articles in trade and enthusiast publications feed directly into what AI systems reference. Timely, topical coverage written by industry publications increases the relevance of your brand in AI responses. For a complete press release template, sample pitch email, and media list building guide, see the Market Your Craft Public Relations Guide.
Participate in Discussion Boards
AI models rely heavily on community platforms and enthusiast forums to gauge sentiment, experience, and credibility. Brands that participate authentically in those spaces build a presence that carries weight in AI recommendations. This is by design: community sentiment on third-party sites is harder to manufacture than a well-written homepage.
The good news is that most craft beverage producers have a built-in community of fans and guests who are already talking about them on these platforms. There is an important distinction between platforms where brand participation adds value and platforms where it can backfire. The guiding principle: brands show up best on platforms designed for brand presence. On platforms designed for community conversation, overly promotional participation reads as noise and can damage the authentic signal those platforms provide. Learn more about best practices for participating in forums specific to your tasting room in our Answer Engine Optimization (AEO) Guide.
As always, we continue to try to stay on top of developments in AI so your team can either establish or support your own position on the technology: whether that’s showing up in AI search or the use of AI productivity tools in your business. We see the benefit of judicious use of AI in craft beverage, otherwise we wouldn’t be spending time talking about it. However, we know it’s not for everyone, which is why looking at both sides of the argument is important. Have an idea or a concern you want to share with the rest of the community? Don’t hesitate to reach out to us at scott@marketyourcraft.com.
We are proud to be speaking at the Craft Beer Professionals Fall Virtual Conference (October 5-6). Andrew Coplon makes it easy for you to pick and choose the virtual sessions that make sense for your beer business, with timely and topical conversations around brewing technique, service and hospitality, data and analytics, and ways to win. Market Your Craft will be presenting From Options To Answers: How AI Drives Traffic To Your Tasting Room on Tuesday at 12PM ET. You can view the complete schedule here. The virtual conference is 100% free, accessible to all, and available forever.
MailChimp archive: https://mailchi.mp/f6d5d2f3de38/260929_thirdparty?e=bd76eedb35
Download: https://app.box.com/s/xq42zv0eb4tg1vm7of2gvn87qncsg3mk
