How Bayer Builds Trust in an AI World | Upside #89

Strategy

8 Jun

How Brands Build Trust Online with Bayer

Your customer has never met you, can't touch the product, and is surrounded by thousands of brands shouting the same thing. Dr. Kishan Rees and Cormac Willer, who build trust in content for one of the world's largest pharmaceutical companies, break down what that actually takes in an industry where getting it wrong has real consequences.

In this episode: Dr. Kishan Rees and Cormac Willer from Bayer on building credibility online, in an industry where trust is genuinely life and death.

Key takeaways

Trust is built by rewarding attention with genuinely useful content, not by leaning on a brand's name or logo, generic, one-size-fits-all messaging increasingly fails online.
"Parasocial relationships" (a term from journalist Deb Cohen) explain why a fake or AI-generated authority figure can build real audience trust before selling something unproven, a risk every brand faces now that anyone can look credible online.
Heavily regulated industries can turn compliance into a trust signal rather than a burden: being transparent about why claims must be fair, balanced and accurate becomes part of the story, not something to hide.
AI's real risk in content isn't using it, it's what happens once everyone uses it the same way: high volume with low actual insight. The better use is generating genuine audience questions and engagement, not just more output.
A new category worth tracking: B2LLM (business to large language model), since being cited accurately inside AI chat tools is becoming as important as traditional visibility, and these models have real blind spots when web search is switched off.
Real trust is built person to person, not logo to person, the strongest examples are founder-led or people-led content rather than corporate branding, true across categories from pharma to consumer brands.
A useful content bar: if it's not an 8 or 9 out of 10, don't publish it, since there's already enough mediocre content competing for the same attention.

Timestamps

00:00 — Why trust is harder to build in a noisy digital market
02:05 — Why generic, one-size-fits-all messaging fails
04:52 — Parasocial relationships and the risk of fake authority
08:59 — Turning heavy regulation into a trust advantage
16:56 — The real risk with AI content: volume without insight
26:24 — B2LLM: a new audience category worth tracking
29:47 — Auditing AI chat tools for content blind spots
31:22 — Wrap-up: trust is built with people, not logos

The detail

Why generic messaging fails, and the parasocial risk
Audiences increasingly discount messaging they've heard a thousand times before, if any company in the category could say the same line, it doesn't land. That vacuum gets filled by "parasocial relationships", a term from journalist Deb Cohen describing how people build trust with someone who looks or acts like an authority, real doctor, AI-generated avatar, or otherwise, often right up until that trusted figure sells something with no evidence behind it. The counter isn't more content, it's content specific enough that no competitor could have put their name on it.

Turning regulation into a trust advantage
In a heavily regulated industry, compliance is usually treated as friction, something slowing content down. The alternative framing used here: be transparent about why claims must be fair, balanced and accurate, and make that rigour part of the story rather than hiding it. The comparison is deliberate journalism and broadcast standards, not marketing spin, applied to content that can't legally speak directly to patients about specific treatments but can equip healthcare professionals with the material to do so.

AI's real risk: volume without insight
The obvious use of AI, more content, faster, isn't actually a differentiator since every competitor has access to the same tools. What's more likely is a flood of content with genuinely low insight per piece. The better use is generating content that prompts real audience questions and engagement, and, separately, auditing AI chat tools directly (testing multiple models with web search disabled) to find where they have blind spots or over-rely on general web consensus rather than accurate specifics.

B2LLM: designing for a new kind of audience
Alongside B2B and B2C, there's now a third audience worth designing content for: B2LLM, business to large language model. Being cited accurately inside AI chat tools is becoming as important as being visible in traditional search or media, especially as more members of the public turn to LLMs directly for information that used to come from a search engine or a broadcaster. Related listening: Storytelling for Founders: How to Make People Care, on why specific, human stories outperform generic brand messaging, the same principle driving the trust argument here.