AI Doctors and Patients // BRXND Dispatch vol 80

Plus, decoding Dolphin speech, the PB&J problem, and borrowing tolerances.

You’re getting this email as a subscriber to the BRXND Dispatch, a newsletter at the intersection of marketing and AI. We’re hosting our next BRXND NYC conference on September 18, 2025, and currently looking for sponsors and speakers. If you’re interested in attending, please add your name to the wait list. As space is limited, we’re prioritizing attendees from brands first. If you work at an agency, just bring a client along—please contact us if you want to arrange this.

What's On My Mind This Week (Noah)

For a long time I've wanted to try my hand at soldering. I don’t have a particularly good reason for it, but I’ve had a handful of occasions where a bit of solder was the only thing holding me back from tackling a project I wanted to do. So when my daughter broke a favorite earring a week ago, I took the opportunity to finally buy myself a soldering iron with the excuse of jewelry repair. Of course, when it arrived I had no idea what to do with it, so I did what I always do these days in such moments: turned to ChatGPT. I took a photo of the soldering iron and asked ChatGPT how to get started assuming I knew absolutely nothing about soldering. As it gave me directions, it also mentioned that under no circumstances should I use the solder that came with the kit, which contained lead, on anything that would touch my daughter’s skin. (It also suggested a silver-based jeweler’s solder and found it on Amazon for me to buy.)

In a way this was just another moment in a long string of moments when AI gave me the confidence to venture into worlds that have previously intimidated me. I’ve felt this sensation over and over again with code, where AI has given me the confidence to try to build with tools and languages I’ve never even heard of, but this was a fun example of that same feeling translating over to the physical world.

So when people talk with certainty about how AI is shrinking the world of creative possibilities, I fundamentally don’t understand. To me what’s been most amazing about the last three or four years has been the expansion of my capabilities.

What Caught My Eye This Week (Luke)

There’s been a lot of renewed excitement lately about AI’s impact on medicine—from improving diagnostics to revolutionizing drug development. This week, the CEO of DeepMind went as far as to claim that the end of disease is “within reach” in the next decade or so. I’m not qualified to assess such claims, but what I do know is that you don’t need to chase moonshots to see AI already improving outcomes in healthcare.

This recent Bloomberg article is a good case in point—highlighting how doctors and patients alike are turning to chatbots for counseling. In the U.S. healthcare system, where primary care visits average less than 20 minutes, and doctors spend as much time on electronic health records as with patients, AI is stepping in to provide something unexpected: empathy.

In studies, many people report chatbots actually have a better bedside manner than real physicians. And doctors are finding value in using tools like ChatGPT to rehearse difficult conversations with patients. As one Stanford physician and researcher put it: “This weird human-computer interaction I’m having is allowing me to practice a high-stakes conversation in a low-stakes environment. Ironically, I think it helped me improve the most human skills I need to be a good doctor.” It’s a good example of how AI can remove the repetitive, time-consuming tasks that prevent experts from doing what they do best.

Will chatbot-based counseling services usher in a future free from pain and illness that AI boosters in healthcare imagine? Of course not—and for that reason, they’re easy to overlook. But time and again, I think the public tends to miss the quiet, incremental improvements AI is already delivering in favor of grand, sweeping narratives.

It’s fine—good even—to offer people an exciting vision to latch onto (this is a marketing newsletter, after all). But I do sometimes wonder: is there a cost to making massive claims like “AI will soon end disease?”

Google had a banner week in terms of launches, including Gemini 2.5 Flash (a faster, cheaper version of Gemini 2.5 Pro) and Veo 2 (a state-of-the-art video model). Oh, and it also released a new LLM aimed at decoding how dolphins communicate.

Nevertheless, Google still has a lot of catching up to do in terms of users…

Coding with AI and the PB&J problem: “If your ‘sandwich’ is a product that doesn’t have an obvious recipe—a novel app, an unfamiliar UX, or a unique set of features—LLMs struggle. They’re great at copying what’s been done before, remixing code that already exists. But ask for something new? Something creative, or specific to your vision? Something that ‘just works’ for a specific use-case? Now you’re back to giving vague or ambiguous instructions to a junior developer from across the globe, who doesn’t know your customer, your context, or what ‘done right’ actually looks like.”

Anthropic shared a list of best practices for agentic coding with Claude Code. The company also released a research tool and Google Workspace integration this week. The latter feature allows Claude to connect directly to your Gmail, Google Calendar, and Google Docs, a major upgrade for the AI assistant.

A Google VP discusses how AI Overviews are changing the entire ad revenue model for search.

The Washington Post is teaming up with OpenAI to integrate its links and reporting into ChatGPT’s answers. WaPo’s willingness to partner with AI players marks a stark contrast with the NYT, which is currently fighting a copyright lawsuit against OpenAI.

The Wikimedia Foundation will give AI developers access to structured Wikipedia data via Kaggle to dissuade scrapers. Welcome to the new era of AI “ergonomics”...

Using AI does not disqualify a movie from winning an Oscar, according to new rules set forth by the Academy of Motion Pictures and Sciences.

AI-generated music now accounts for 18% of all tracks uploaded to Deezer.

Perplexity is in talks to integrate its AI voice assistant into Samsung phones.

OpenAI has launched a new Flex processing option that cuts API costs in half for the o3 and o4-mini models, offering a more affordable alternative with slower response times.

On borrowing tolerances and AI integration (from the Alephic blog): “AI works best when it's just one piece of a larger technical puzzle—borrowing the tolerances of its adjacent parts. Thoughtful database structures, careful API integrations, smart application logic, and intuitive user interfaces can enhance the AI-powered service within, providing context to the larger system.”

Sama: Please stop saying please and thank you to ChatGPT.

---

If you have any questions, please be in touch. If you are interested in sponsoring BRXND NYC, reach out, and we’ll send you the details.

Thanks for reading,

Noah and Luke