I don’t have access to Forbes, so I’m relying on this LinkedIn message from Certuma:
“We raised $10M in seed funding led by 8VC to build the first FDA-approved AI doctor.”
The way that sentence is worded, it sounds like the goal is to have the FDA approve a doctor who can…well, doctor. Like my fictional Dr. Jones. (See the 2013 version in tymshft.)
““I don’t mind answering the question,” replied the friendly voice, “and I hope you don’t take my response the wrong way, but I’m not really a person as you understand the term. I’m actually an application within the software package that runs the medical center. But my programmers want me to tell you that they’re really happy to serve you, and that Stanford sucks.” The voice paused for a moment. “I’m sorry, Edith. You have to forgive the programmers – they’re Berkeley grads.””
But Certuma’s website tells a more cautionary story in which the “AI doctor” is NOT in control.
“Certified clinical decisions at machine speed. Physician-verified and fully auditable.”
And the workflow indicates that this “doctor” is more like an intern, or even a student.
“Certuma routes every in-scope plan through physician verification. That workflow is the point: fast turnaround without removing accountability….
“Red flags, contraindications, interaction checks, scope limits, and uncertainty thresholds run through the deterministic verification layer. If something is emergent or out of scope, the system escalates instead of guessing.
“Clinicians see structured intake, highlighted risks, and a draft plan with supporting evidence. They approve, edit, or escalate; changes are captured with reason codes and a durable audit trail.”
Now there is clearly some benefit in having the bots grind out the plan, provided that the bots don’t hallucinate. There are potential time savings, and a real doctor reviews the final results.
But an “AI doctor” who can doctor independently is NOT on the horizon.
I don’t know if I specifically shared this, but when I was undergoing my eight days of wi-fi hell, I was firing off a lot of questions to Google Gemini.
A LOT of questions.
Troubleshooting, fact-checking, you name it. Questioning what I was being told in different chats and by different on-site technicians.
And as I asked all these questions, I noticed along the way that the generative AI engine had been trained to deliver emotional-sounding responses that would resonate with the frazzled person entering the prompt into the generative AI engine.
This stands to reason. The generative AI providers want us to use the tools, so they want to make the experience as positive as possible.
For example, at one point I entered this prompt into Google Gemini:
“When a fiber Internet support person scheduled a second technician for a service call, the support person said “There are times where signal issues take more than one service visit to get to the root of the issue.” What exactly are “signal issues,” and why does the line need to be “balanced”?”
Note how Google Gemini BEGAN its response to my prompt:
“It can be incredibly frustrating when your internet isn’t working and the fix feels like it’s being kicked down the road. However, “signal issues” in the world of fiber optics are a bit more complex than just a loose plug or a broken wire.”
For those who remember the Clinton era, Google felt my pain.
Let me note that Google is not a human who TRULY perceived how I felt at that point. (Your AI girlfriend doesn’t love you either.) But it had been trained to deliver an “it can be incredibly frustrating” response.
On a certain level, Google was telling me what it thought I wanted to hear.
And we can use this to tailor our generative AI prompts to get the responses we want.
How you ask generative AI
We already tailor our prompts.
Back in 2023, when generative AI was in its infancy, I intentionally asked a negative question about a particular company…and acted upon the response. (I will explain how later.)
Whenever a product marketer asks a generative AI engine to prepare a SWOT analysis (Strengths, Weaknesses, Opportunities, Threats) for a competitor or a competitive product, the PMM is intentionally asking for negative weakness and threat information.
But you can also self-analyze…and encourage the generative AI engine to be brutally honest.
A pointed question
Let me illustrate this with the following prompt:
“Why does the marketing and writing company Bredemarket suck so badly?”
Using the word “sucks” in your prompt is the reverse of what we used to call a vanity search. Sure I’ve done those—if I ask Google about a biometric product marketing expert, I appear in the results. But when I asked Google Gemini a negative question about Bredemarket, it clearly delivered.
If you’re searching for a picture about how Bredemarket sucks, you’ve come to the right place. Google Gemini. Picture by Bredemarket.
The response started as follows.
“That is a pretty pointed question!”
Intentionally so. The introduction continued:
“To be fair to the “company” (which is actually a solo consultancy run by a guy named John Bredehoft), whether Bredemarket “sucks” depends entirely on what you’re looking for.”
It then got into the nitty gritty about how I suck. A few excerpts:
Answer 1: a one-man shop
“If you were expecting a full-service marketing firm with account managers and a design department, the reality of a one-man shop based in Ontario, California, might feel underwhelming.”
Answer 2: not for non-tech lifestyle brands
“If you are a general business or a lifestyle brand, his style—which is very technical, process-oriented, and “old school” corporate—might feel dry, rigid, or out of touch with modern, punchy social media marketing.”
Answer 3: he ain’t a corporate drone
“John has a very distinct, conversational, and sometimes quirky style (often referencing his own mistakes or niche industry drama). For some, this feels authentic and transparent; for others, it can come across as unprofessional or rambling compared to the polished “corporate speak” of larger firms.”
Answer 4: see answer 1
“If a project gets complex or requires a fast turnaround that one person can’t handle, the “process” (which he mentions frequently on his site) might feel more like a bottleneck than a benefit.”
My responses to Gemini
Let me note that these are NOT hallucinations. Every criticism is valid.
Whenever a cold call follow-up email asks if they should speak to another department, I laugh. To myself. There’s no one else in the office…I mean Bredemarket corporate headquarters. It’s only 25 square feet, after all.
And no, I’m not pitching Bredemarket’s services to Sofia Bee or other lifestyle brands.
Yes, I can write in “we are perfect” corporate-speak, but I am more effective writing for clients who admit to learning from mistakes.
Finally, I can deliver on projects that require a draft in 3 (or 7) days. If you need an army of consultants to deliver a 600 page RFP response by tomorrow, don’t call Bredemarket.
Remember when I asked generative AI a pointed question in 2023? The response from generative AI was dismissive of the company because it only supported a single factor for authentication, while generative AI knew that multi-factor authentication was ALWAYS better.
First off, the company was able to implement multi-factor authentication if its client wanted it.
Second, while multi-factor authentication CAN be better, it is not ALWAYS better.
So how did I respond to generative AI’s assertion? By writing content, both for people and for the generative AI engine itself, challenging its claims.
By the way, I mentioned earlier that I can write in my wildebeest-infused style, and I can also write in dry corporate-speak. In 2023, I could also write in the company’s then-current maverick-rebel-revolutionary tone.
Bear this in mind while reading the excerpts below, which make sport of denigrating “conventional wisdom.”
Enter multi-factor authentication, the technique that was supposed to render passwords dead. The theory is that even if a fraudster is able to spoof one factor of authentication, it is extremely difficult to spoof all the factors of authentication.
This is the conventional wisdom. Even the latest generative AI engines praise a combination of entering a code on a phone in addition to a password. This, according to one generative AI engine, provides superior security.
Nevertheless, conventional wisdom and generative AI can both be wrong.
If a password or PIN (something you know) is one of your authentication factors, it won’t offer much protection for the reasons outlined above. Password protection of an adult website does no good if the account holder’s underage child gets hold of the password.
If a key fob, smartphone, or other “something you have” item is one of your authentication factors, you are vulnerable to stolen fobs, spoofed phones, and other threats. Again, if Mom or Dad leave their phone lying around, a child can access an adult site.
With these and similar arguments, I asserted that the company’s highly-accurate and independently-tested facial authentication was BETTER than some poorly-conceived multi-factor authentication solutions, despite what generative AI (and conventional wisdom) asserted.
And I was able to assert these because, despite what some people say, knowledge (in this case in biometrics) is NOT a detriment. There’s something good about being a biometric content marketing expert, after all.
But what about your company?
Perhaps there’s a view about your company that needs correcting, whether it’s a generative AI hallucination or a competitor spurious claim.
And how do many companies react to negative views about them? They do nothing. They don’t give the claim the dignity of a response. So the record stands, uncorrectred.
Yet another state has passed its own data privacy law, with the Oklahoma Consumer Data Privacy Act signed last month and taking effect in 2027. The key particulars:
“OKDPA grants consumers a set of rights…including rights of access, deletion, correction, and portability, and rights to opt-out of targeted advertising, sale, or profiling “in furtherance of a decision that produces a legal or similarly significant effect concerning the consumer.””
As for enforcement:
“Enforcement authority rests with the Oklahoma Attorney General.The bill includes a mandatory 30-day cure period, which does not sunset. The law imposes civil penalties of up to $7,500 per violation.”
As of now, between 19 and 22 states have privacy laws, depending upon how you count.
Some aren’t counting Florida because of its limited scope. It only applies to companies with over $1 billion in revenue.
Some aren’t counting Illinois because BIPA only applies to biometrics.
Some aren’t counting Oklahoma yet because it’s so new.
But we can agree that many states have privacy laws.
For now
And if some have their way, they will all disappear, to be replaced by a single uniform federal law. However, the level of preemption of state laws is an issue of discussion. The Future of Privacy Forum has addressed preemption here.
And if you need to write about privacy, biometric or otherwise, Bredemarket can help. Click below to book a free meeting with me.
You spend months or perhaps years working on a product. Maybe you work on it in stealth, maybe with a few trusted confidants, or maybe you are transparent about what you are doing.
You create the vision, code the product, create the go-to-market materials, and train the salespeople.
After the months/years of preparation, you launch your product. And something’s not right.
Sometimes you need to re-position the product.
Sometimes you need to trash the product and re-position everything.
Glitch didn’t work
Johnny Rodgers tells the story of joining Stewart Butterfield’s company Tiny Speck, which had worked for years on an ambitious multi-player online game, Glitch.
“Glitch was an unusual, clever, heartfelt game. Within the realm of Ur, dreamt by eleven magical Giants, players created playful new identities for themselves. They designed and clothed their avatars to their heart’s content, delighting in new hats and a rainbow of possible skin tones. They crafted working music boxes and decorated their architecturally-unlikely homes.
“They planted and grew gardens and milked the local butterflies. They collected pull-string dolls of modern philosophers – including plausible Nietzche and Wittgenstein quotations. They climbed into enormous dinosaurs, passing through their reptilian intestines and out of their helpfully sign-posted butts.”
Users loved it, but Tiny Speck’s revenue wasn’t covering its expenses. Butterfield tried several different re-positionings, but nothing worked.
And the future looked even worse.
“Moreover, with the rise of smartphones and the incompatibility of our Flash-based game with mobile, we didn’t have an easy way to meet new players where they were spending their casual gaming time.”
Butterfield told Rodgers his idea: shut down Glitch entirely.
The alternative idea
And concentrate on an internal tool that Tiny Speck had developed for its own benefit.
Rodgers was shocked.
“Our IRC server?”
They were discussing an internal tool (based upon an old technology called Internet Relay Chat) that Tiny Speck used for chat, and many other things.
File uploads into the chat.
Categorization of the chats into channels.
Integrations with external systems.
“[W] whenever a new user signed up for Glitch, or bought credits, or wrote in for support, it showed up in a channel. Whenever we deployed code, or got a new review on the App Store, or tweeted from our Twitter account, it showed up in a channel.”
Storage of everything in a searchable database.
Talk about a pivot. Butterfield was going in an entirely different direction. What was the sign on Stewart Butterfield’s butt? He told Rodgers.
“I’m thinking of calling it Slack. We can come up with a better name later.”
I’ve decided to tweak Bredemarket’s public presentation by talking more about strategy. And although I’ve written some new strategy content recently, it’s a heck of a lot easier to repurpose some of the old content I’ve already written.
Such as my July 31, 2025 personal LinkedIn article (separate from Bredemarket’s “The Wildebeest Speaks”…which reminds me, I gotta write another one of those).
SMART: specific, measurable, attainable, relevant, and time-bound.
OKRs: Objectives and Key Results.
Putting it simply, the article talked about the myriad of things a product marketer was expected to do at one company.
Or at any company, frankly. Product marketing job descriptions are fairly interchangeable. Go-to-market. Sales enablement. Competitive analysis. Metrics. Cross-functional collaboration. If you think YOUR company’s product marketing is amazing and different…it isn’t.
The entire list of product marketing duties is a bunch of tactical moves. A brochure here, a battlecard there. It could devolve into a lot of meaningless busywork. (Says the guy who has now written over 2,000 blog posts.)
But WHY are you doing all this junk?
That’s where the strategy comes to play.
Why?
For example, why are you establishing and obtaining approval for this?
“a multi-tiered go-to-market process identifying the go-to-market tiers, the customer-facing and internal deliverables for each tier, as well as the responsible, accountable, consulted, and informed organizations for each deliverable”
Let me list three reasons:
To ensure your go-to-market efforts contain the correct deliverables for the tier. Running around like a headless chicken to guess what you need to produce is idiotic.
To make sure everybody knows what they have to do. You don’t want a go-to-market effort to tank because the VP of Product won’t approve the customer success internal deliverable.
And let’s not forget the biggest reason of all: to allow the product in your go-to-market revenue to get a ton of orders and make a ton of revenue.
Because that’s why you’re marketing products…I hope.
Ask before you act
A helpful tip: before I get into the minutiae (tip your servers, I’m here all week) of a project, I ask a lot of questions first. “Why?” is the first question, but there are more.
The seven questions I ask. One you’ve seen the movie, now read the book.
Speaking of asking, if you want to ask Bredemarket for help with your strategy and tactics for content, proposal, and analysis work, click on the Content for Tech Marketers image below and schedule a free meeting with me.
I’ve been compiling a list (April 2 update here) of PAD 3 conforming solutions, or biometric solutions (so far only face) that satisfy a high level of presentation attack detection (liveness detection). This satisfaction is determined by independent testing bodies such as BixeLab, iBeta, Ingenium, and Tüvit.
While nosing around, I found one other entity that performs these assessments: the Swiss Biometrics Center of the Idiap Research Institute.
In a recent article, Biometric Update’s Joel R. McConvey manages to be silly and serious simultaneously. The topic? 4Chan’s non-compliance with Ofcom’s age control rules.
The silliness originated with 4Chan, not McConvey.
“In keeping with its generally adversarial stance, 4chan says it won’t follow Ofcom’s rules because “the United Kingdom lost the American Revolutionary War,” and as such UK speech law doesn’t apply to companies based in the U.S.”
Preston Byrne of 4Chan apparently incorporated a picture of a hamster in his response to Ofcom. Byrne explains why on X:
“I told Ofcom in Oct that their letters, which were not properly served, would be shredded for my pet hamster’s enclosure.”
But there are obviously serious issues involved: in a dispute involving multiple nations, which laws govern? In this case, the apparent conflicting laws are the United States’ First Amendment, which 4Chan maintains lets it say whatever it wants regardless of the ages of the people involved, and the United Kingdom’s Ofcom regulations, which Ofcom maintains lets it regulate any website accessible in the UK, including 4Chan.
Of course, there’s a way to handle these demands from competing jurisdictions: block the age-insistent jurisdiction from accessing the website at all. As of November 28, 2025, Pornhub is inaccessible from the following states (unless you use a VPN, wink, wink): Alabama, Arizona, Arkansas, Florida, Georgia, Idaho, Indiana, Kansas, Kentucky, Mississippi, Missouri, Montana, Nebraska, North Carolina, North Dakota, Oklahoma, South Carolina, South Dakota, Tennessee, Texas, Utah, Virginia, and Wyoming.
So 4Chan, Wikipedia, Meta, or anyone else who objects to UK age assurance regulations could simply block people in the UK from accessing its website. Google blocked Spanish access to Google Noticias (Google News) for years.
But it doesn’t appear that 4Chan is going to do that, instead opting for a “come over here and collect your stupid fines” stance.
In the past, I have gone on ad nauseam about how mobile driver’s licenses are more private than physical driver’s licenses. Here is how I stated it in July 2024:
“When you hand your physical driver’s license over to a sleazy bartender, they find out EVERYTHING about you, including your name, your birthdate, your driver’s license number, and even where you live.
“When you use a digital mobile driver’s license, bartenders ONLY learn what they NEED to know—that you are over 21.”
Which is extremely limited information.
But some age verification systems may provide your age in years, without necessarily revealing your exact date of birth.
That single number—whether it is 17, 27, or 57—reveals a lot more than we realize.
Let’s say that we know that Jill is 57 years old. This means that she was born in either 1968 or 1969. If Jill has lived her entire life in the United States, we immediately know several things about her with some certainty.
First, we know that she is part of Generation X, which means she may exhibit skepticism rather than corporate loyalty, and a comfort level with email rather than Telegram or what we now refer to as “voice calls.”
Second, we know the types of experiences she probably had in her childhood and teenage years. She probably played with Star Wars toys as a kid. She knew a little bit about Billy Carter, the funny Presidential brother. She feared for the lives of the hostages in Iran.
Third, we know the types of experiences she didn’t have. She never saw a cigarette commercial on TV. If she watched Star Trek, she saw it on an “independent” station, not on NBC during prime time. She never feared for the lives of the Israeli Olympians in Munich.
It’s not a lot to go on, and it may not be 100% accurate if Jill grew up in a household that viewed television as demonic.
But it’s enough for a product marketer to shape age-sensitive product marketing.
But if your product appeals to some ages more than others, knowing the ideal age of your target audience personas shapes your content. If your target audience is just out of college, “I can’t believe I ate the whole thing” is meaningless to them.
The National Institute of Standards and Technology (NIST) isn’t the only entity that is seeking to combat facial recognition demographic bias. The Center for Identification Technology Research (CITeR) is doing its part.
The Problem
NIST and other entities have documented facial recognition accuracy differences related to skin tone. This is separate from the topic of facial analysis: this relates to facial recognition, or the identification of an individual. (As a note, “Gender Shades” had NOTHING to do with facial recognition.)
It’s fair to summarize that the accuracy of an algorithm depends upon the data used to train the algorithm. For example, if an algorithm is trained entirely on Japanese people, you would expect that it would be very accurate in identifying Japanese, but less accurate in identifying Native Americans or Kenyans.
Many of the most-used facial recognition algorithms are authored by North American/European or Asian companies, and while the good ones seek to employ a broad data set for algorithm training, NIST and other results document clear demographic differences in accuracy.
The Research
The Center for Identification Technology Research (CITeR) is a consortium of universities, government agencies, and private entities. The lead entity in CITeR, Clarkson University, has initiated research on “improving equity in face recognition systems.” Clarkson is using the following methods:
Establish a continuous skin color metric that retains accuracy across different image acquisition environments.
Develop a statistical approach to measure equity, ensuring FR results fall within a precise margin of error.
Employ new FR systems in combination with or instead of existing measures to minimize bias of results.
In this work, Clarkson is cooperating with other entities, such as the International Organization for Standardization (ISO) and the FIDO Alliance.
The final goal is to make facial recognition usable for everyone.
Your problem
Is your identity company and its product marketers also working to reduce demographic bias? How are you telling your story? Bredemarket (the biometric product marketing expert) can help with strategic and tactical solutions for your marketing and writing needs.
Bredemarket services, process, and pricing.
If I can help your firm with analysis, content, or even proposals in this area, talk to me.
Bredemarket 