Entries

A friend of the list shared this excellent perspective with me the other day: “AI isn’t the app, it’s the UI”.

It’s a great perspective on how to think about generative AI with respect to a product or productized services business.

(Sidebar: The title of the article alone lets you start to unpack mentally before you dig into it. I love that quality in a title).

If you could make an article heat map, this sentence would be a fire-red:

The startups that come out on top of the AI hype wave will be those that understand generative AI’s place in the world: not just catnip for venture capitalists and early adopters, not a cheap full-service replacement for human writers and artists, and certainly not a shortcut to mission-critical code, but something even more interesting: an adaptive interface between chaotic real-world problems and secure, well-architected technical solutions. 

The chaotic real world problems – that’s what we bring. We also bring chaotic solutions embedded in our natural way of speaking, which proceeds from our equallty chaotic way of thinking, as Stephen Wolfram has been hammering on in his recent podcast tour.

When an LLM is created, we expose it to enormous amounts of our chaotic minds.

In some cases (ChatGPT for instance, but not Claude), we also expose it to our public attempts to control technical solutions, like  web programming languages or operating systems. Its understanding of how we do that is where the UI opportunity lies: “adaptive interfaces”.

Imagine a hammer that changes shape based on the hand that holds it; that’s the UI that we have the potential of crafting, except instead of a hammer, it’s just words that change.. or is it?

A person who’s charming has the ability to talk to you and make both of you feel as though you’re wonderful

– Drayton Bird, here

Question for you: what happens when you love using a product, and you sense that whoever made the product also loved making it?

One answer: charming happens.

I think this is why we see products saying, “Made with ❤️ in Ulaanbaatar”. (Or any city). The intent is to infuse the product with some positive feeling and convey it from maker to user.

And there you have it, a definition of a product which is charming: creating a positive feeling in the user while giving them the feeling that the makers of the product shared in that positive feeling.

A bit clunky but hopefully get the idea.

But these “Made with ❤️” messages don’t pack as much of a punch now; they feel automatic. Is there a new to be charming in your product messaging?

 

 

One interpretation of the movie, The Black Swan, is that the lead character, the Swan Queen and her rival, the Black Swan, are two aspects the same person.

Like Jung’s “shadow self”, the repressed black swan is a problem – a tangle of motivations and desires that are not acknowledged or even understood.

Is our refusal to acknowledge them a matter of self-survival? Ogilvy thought so. He said: customers fear that accepting their hidden desires will destroy their vision of themselves – and like a protective blanket they need that vision to survive.

But why? Because they fear that the desires will run wild; that was the Swan Queen’s fear. Examples of black swans running wild:

• Extreme laziness
• Extreme overeating and drinking
• Obsession with material things and money
• Controlling behavior

But repressing them doesn’t mean they don’t exist; that’s why fine-tuned messaging speaks to the Black Swan and says it’s Ok.

Netflix: it’s delightfully ironic to binge-watch series all weekend!

Here’s the thing, though – “swan-whispering” isn’t always sinister, because in moderation, most desires are harmless if not fulfilling. Using Basecamp and Asana, the team gets some control over its work – that’s nice for everyone.

There’s a product design and messaging sweet spot here.

The Checklist Manifesto isn’t about checklists. It’s about managing extreme complexity among multiple actors.

There other ways to deal with this issue – Agile, for example.

But checklists compress the extreme complexity challenge into a smaller, more mentally manageable space. And they make it easier to communicate it.

Just to ground this commentary, examples of such complexity are:

• The end to end behavior of a multi-step form
• A SaaS product roadmap, with growth, brand, and product goals balanced
• Acquiring and assimilating not just a product and its tech stack, but a brand and its workforce

One of the greatest complexities, though, has always been: “how does the world’s leading political, economic, and cultural power present itself to the world?” A lot of work goes into this and a lot of money too.

In other words, one of the greatest complexities has always been messaging

Modern governments struggle with this now, just as ancient ones did, like Babylon nearly 3000 years ago.

We saw this yesterday in looking at the messaging embedded in their map:

• Babylon sits at the intersection of the world
• Babylon is a great commercial hub and a center of riches
• Babylon is a cosmopolitan city of knowledge, learning, artistry
• Babylon is the most multicultural and multi-linguistic place in the world – you are welcome here

It’s ambitious, magical, it distills extreme complexity into the most manageable pieces. It’s easy to convey. Many different people can “work on it” in collaboration.

It hews perfectly to the spirit of The Checklist Manifesto.

The funny thing is though – at the end of the day it’s just a list.

 

There was a guy in Sicily who obtained high quality prints of historical maps, hand-colored them, and sold them in the public market in Catania, not far from the seafood stalls.

When he sold them, he described the historical and cultural details they captured in detail. He was a cultural historian; the lecture was part of the deal for me.

Was he a “map maker”? I’m inclined to say yes, but wth do I know.

Message Maps isn’t a “real” map maker either; instead, its “strategic message maps” exist to help grow a business, basically.

Meanwhile, the first map of any kind, as far as we know, was Imago Mundi, created by a Babylonian 2,700 or so years ago.

Here it is with Babylon at the central point of the world’s great commercial and cultural internet, the Euphrates (the vertical swath):

Nowadays, geographic maps are more utilitarian but this map’s main job was not to “give you directions to a taco stand” but to convey key messages:

• Babylon sits at the intersection of the world
• Babylon is a great commercial hub and a center of riches
• Babylon is a cosmopolitan city of knowledge, learning, artistry
• Babylon is the most multicultural and multi-linguistic place in the world – you are welcome here

So I guess I’m not the first to make a message map.

But here’s the real takeaway: strategic message maps are nothing new and they have helped organizations grow for a long time.

“Build uncool solutions for uncool problems”

– Anonymous

You’re kitchen sink has a leak – what do you do?

For most of plumbed history, you called the plumber. But now, you use plumber.ai.

Plumber.ai lets you diagnose the problem with your phone using pictures, video, audio, and a series of questions about about the particulars of the problem. It works something like this:

• It analyzes the imagery you provide using computer vision to find useful clues
• Each question it asks you is newly crafted with generative AI based on everything it has learned up to that point
• Based on the above steps, it advises you on what to do next: fix the problem yourself, which will involve ordering replacement parts for you, or call a plumber.

When you “call” a plumber though, the plumber also diagnoses the problem remotely – also using AI, repeating a similar process.

*      *      *

As of now, plumber.ai is a non-existent business (and Handyman Hank is a long way off of this vision). Will it exist? I have no idea. Will it belong to reddit, because reddit owns the data? I don’t know.

I’m just trying to make a point: in the near-term future, the imperative for many solutions providers, especially if you can get your hands on the right data, will be to to make an AI version of whatever human best understands the solution.

I was talking to a friend today about small things, such as smart cars, folding fans, and “bets” – as in the small actions you take in business and product development.

It got me thinking about product messaging – and keeping it short.

You might have heard the short story by Margaret Atwood, who wrote A Handmaid’s Tale:

“Longed for him. Got him. Shit.”

The “him” in this story could have authored this short story:

“I met my soulmate. She didn’t.”

Or maybe you’ve heard this bullet of sadness attributed to Hemingway:

“For sale: baby shoes, never worn.”

The last hits hardest, but but they’re all good; they all hook you in and tell the story at the same time.

At a campfire or a kitchen table – the skillful storyteller captures the entire essence of a story in just a few words, hooks in their listener, assesses  interest, then decides how much of the rest of the story to tell.

As it it with a case study – get good at compressing them down to one sentence and they become hooks.

The point is: don’t think you don’t have enough room to tell a story in your messaging.

When you work at a huge software firm, there’s a behind-the-veil effect when you realize this: 1-to-1, non-scalable relationships are a massive part of the business model.

Every such firm has its own services, some free, some paid,  from customer support, technical services, and training all the way up to full-service digital agency and consulting services. All provided by full-time humans, not code.

On the same token, large consultancies and agencies package more IP with their services, the bigger they get. Sometimes even a micro-SaaS.

Conversely, if a software company is smaller, there are usually fewer services around it, maybe just customer support. And when a consulting firm is small, there isn’t much IP.

Thus, most small firms in the B2B space are one or the other – solutions or services. With this distinction come very different company structures, sales and marketing processes, and brand messaging.

But LLMs might change that.

Small services firms can quickly build custom software products for clients – if they use LLM assistance in research, coding, system administration, UX design, and data management.

Small product firms can create  bigger customer support,  account management, and client services programs – if they leverage LLM capacities in research, content creation, sentiment analysis, data science, and categorization.

In either case, it’d be humans leading this work, of course. But a special kind of human – ones who know how to ask their LLMs the right questions.

What if you had an add-on for your product that let it do:

1. general research and documentation
2. summarization
3. categorization
4. sentiment analysis
5. personality analysis
6. code generation/editing
7. configuration assistance
8. pattern matching/vetting
9. editorial assistance
10. project management
11. questioning & ideation

Well you do, you just have to figure out the right way to add it.

To do so, it is crucial to become good at writing custom add-on code – ie. mastering your human language of choice. Which in turn involves mastering the art of providing a series of questions in a sequence, with context as needed, like Socrates, any artful purveyor of the 5 Whys, and surely, our homo sapien ancestors who first started to develop modern natural language technology 50k to 100k years ago.

Anyway, you guessed it – the add-on is LLM technology and the proof of concept is ChatGPT.

Jonathan’s right: Steve Job’s “next big thing” is here.

And yes, I left content generation out on purpose –  that’s better done by humans, at least if you’re interesting in getting new ideas out of your writing, and sparking them in your readers.

So what do mr. clean, pringles, and quaker oats have in common with b2b brands like mailchimp, hootsuite, and intercom?

Faces  – they all have faces.

We like faces by genetic inclination; a University of Nottingham study from 2015 showed that people prefer a gambling machine with a human-esque face.

But why do we like faces? A subset of gamblers surveyed who believed they had the ability to manipulate people were more likely to choose the humanoid one-armed bandit.

In short, we like faces on our brand because we like people, even in our business software, we want there to be someone there. Like Clippy but not an annoying geek.

Clippy was sort of a starting point for anthropomorphism in products and while he was “fired”, he’s back.

If you trace the arc of product UX since the advent of Internet software, the human presence gradually inserts itself.

For example, product onboarding went from “read the manual please” to a guided human-like tour, then  we got personalized recommendations, then chatbots.

B2B tech platforms levelled up in the last decade by pairing chatbots with not just real people, with clear faces and real names, but with intelligent people; for a time, a support chat on Pantheon.io yielded a conversation with legit open source web developer.

But the pinnacle is the real person you know, collaborating inside the software with you, inside the tool, doing the work with you – Basecamp, Balsamiq, Invision, Figma, Google Docs.

That’s quite an arc – from a the mustachioed cartoon guy on the Pringles can, all the way to the real person you know and trust, in the chat thread.

The message is pretty simple: use our product and work in collaboration to get what you need.