Each week we'll gather headlines and tips to keep you current with how generative AI affects PR and the world at large. If you have ideas on how to improve the newsletter, let us know!

What You Should Know 

AI Search Changes What a Digital Front Door Should Look Like

People have spent decades finding brands through search results. They typed a question into Google, scanned the links, and decided which website looked most useful. AI search changes that behavior. The answer arrives without a click, and a brand may only be considered if it appears in the response.

Google is pushing that experience deeper into the browser. The company recently announced updates to AI Mode in Chrome that let people browse webpages side by side with AI, ask follow-up questions without leaving the page, and add recent tabs, images, and files as search context. 

ChatGPT is also influencing research and decisions. OpenAI said in February that ChatGPT reached 900 million weekly active users, and its report last fall said 49% of messages were categorized as “asking” prompts. While traffic from ChatGPT is still relatively low (see above about zero clicks), Semrush found that outbound referral traffic from ChatGPT to the web grew 206% from January 2025 to January 2026.

When someone asks which platform to use, which provider to trust, what product fits a need, or how one company compares with another, AI decides what belongs in the answer. If your brand is missing, misunderstood, or described with stale information, users won’t walk through your digital front door.

So, what should websites look like in 2026? 

A recent Penta white paper says they should be written for AI as well as humans. In one recurring audit, Penta found roughly 60% of cited material came from corporate-owned or business-produced sources, compared with about 25% from media outlets and 15% from user-generated platforms and third-party institutions. The firm also found that AI systems favor owned content that is clearly structured, relevant to questions buyers are asking, and easy to parse, including tiered headings, FAQ-style formats, and regularly updated material.

Many brand websites don’t reflect that. They rely on clever but unclear taglines, broad claims, vague positioning, and “solutions” language that often doesn’t explain to humans, much less AI, what the company actually does. If a page does not clearly explain who the company serves, what problem it solves, what makes it different or better, and what proof supports its claims, the brand is likely invisible in AI search.

That means webpages need to become more useful and specific. They should answer the questions people are likely to ask, use clear category language, include current proof points, explain comparisons fairly, and link to credible third-party sources. Press releases, earned media, executive commentary, analyst materials, and product pages should also tell the same story, because AI systems pull from multiple sources.

Visibility depends on whether a brand is relevant, understandable, credible, and consistent across the web. That requires more than technical optimization or keyword stuffing. Clear messaging, useful content, and substance help AI recognize whether a brand belongs in an answer.

Start with the pages that matter most, like the homepage, product pages, audience pages, location pages, and resource hubs. Ask whether they clearly explain who you are, what you do, who you help, why it matters, and what evidence supports the story. 

Once you have content on those pages that AI can easily parse, you’ll have a digital front door built for the future of search.

Elsewhere …

• LISTEN: HumanX Showed Us Who’s Using AI and Who’s Faking It
• AI Helps Online Ad Businesses Boom
• Which Pothole To Fix? AI Team Helps Company Develop City System
• FDA Turns to AI to Speed Up Clinical Trials

Tips and Tricks

🥔 How to turn “Spud” on

What’s happening: Last Thursday, OpenAI released GPT-5.5, a new model that reportedly went by the internal codename “Spud.” While OpenAI touts benefits in coding, it also says GPT-5.5 is better at knowledge work. 

“Because the model is better at understanding intent,” the company wrote in its blog post, “it can move more naturally through the full loop of knowledge work: finding information, understanding what matters, using tools, checking the output, and turning raw material into something useful.”

However, you have to do a bit of work to turn it on.

Under the hood: ChatGPT’s default is the “instant” model, which is GPT-5.3. To access GPT-5.5, you need to click the dropdown menu and select “thinking.” If you click “configure,” you can also access previous models, but GPT-5.5 has been producing outputs that communicators may resonate more with.

How so?: For weeks, it seemed that GPT-5.4 was more rambly in its responses, with too much filler text that didn’t meaningfully tell a story. It also didn’t have strong research skills and would routinely surface dated material or aggregator websites where it was difficult to identify the original source.

It’s early days, but GPT-5.5 seems to produce shorter, sharper responses that actually answer questions and identify potential problems with research. For instance, one search for statistics about turnover rates in the insurance industry identified that a source was conflating the turnover rate for agents with the turnover rate for adjusters.

Quote of the Week

“The idea of standing out is no longer optional. There’s a real risk of sameness.”

— Andrew Warden, CMO at Semrush, speaking at Adobe Summit last week, where he warned of a “bland tax” because AI is ignoring uninteresting material

How Was This Newsletter?