“The Right It” by Alberto Savoia Book Review, PLUS a Midjourney Pretotyping Experiment

A week or two prior to this posting, I did an AI experiment to pretotype how capable ChatGPT was at writing code and wrote about it on LinkedIn. AI’s ability to code was unsurprising, everyone’s talking about it. But for some (or perhaps many) the concept of a pretotype was — surprising that is.

“Pretotype? Don’t you mean ‘Prototype’ Adam?”

NO. I meant Pretotype. And for those unfamiliar with the concept, it is a term coined by Alberto Savoia in his groundbreaking (yes, really groundbreaking) book “The Right It”. Savoia is the Founder and Innovation Agitator at Pretotype Labs, a former Innovation Lecturer at Stanford, and also the former Engineering Director and Innovation Agitator at Google Labs.

In Safi Bahcall’s book, Loonshots, he talks about the concept of a Product Champion. To summarize, in my opinion, who exactly Alberto Savoia is — you might consider him as an Innovation Champion wherever and whenever innovation is to be done. His is the type of career I aspire to most.

As for the book, and why “groundbreaking.” And perhaps — why groundbreaking NOW moreso than ever before. Well, for many, the idea of a Minimum Viable Product (or perhaps the Minimum Viable Prototype) are often what we think of, and aim for, when pursuing the birth of a new idea or product. The problem is: that many ideas are insanely time consuming to work on… EVEN if you’re only aiming for an MVP. In Savoia’s book, we are reminded and confronted with the idea that time is finite, and whatever thing you are working on also represents the opportunity cost of whatever else you cannot work on.

A mantra you may have heard: “THE IDEA MIGHT BE GREAT, BUT MAY NOT BE GREAT FOR YOU.” is a theme covered in this book. Pretotyping will help you identify not only if an idea is good, but how capable you are to bring it to life.

The Right It introduces you to a methodology of battle testing an idea long before you find yourself trapped in a Sisyphean cycle of trying to bring an MVP to life. You learn the value of doing your own research, and why. And more importantly, how doing your own research (as opposed to depending on studies) can protect you from wandering into “Thought Land”. A place where unicorns (no, not the companies) coexist with Santa Claus, the Tooth Fairy, and your idea is brilliant.

Key points of Pretotyping include testing them fast with the smallest possible investment, to determine if an idea is even worth pursuing. The emphasis is on what you can learn to improve faster. Pretotpying helps you cover ground quickly in order to uncover what the best/most appealing version of your idea is. But, a Pretotype does not exist in a vacuum. It is the thing, a mock-up, a simulation, or even offering (perhaps for something that does not exist) that is just enough to test your assumption that you are on the right track. To be successful with Pretotyping, you’ll want to make sure that your experiment helps you better understand your value proposition, the features, pricing, target market, and demand. You will be most successful with Pretotyping by embracing an experimental mindset, a willingness to be wrong (and recognize it) and the ability to embrace failure as an opportunity to learn and iterate.

OKAY… but, how is this groundbreaking? If you’re familiar with certain business concepts like Agile, or Lean, you may appreciate that you can be iteratively working towards the perfect MVP, and “doing everything right”, only to discover… you never should’ve began in the first place… that the idea itself was wrong. What’s more is the emphasis on doing your own research. While some may frown on you doing your own research: you don’t have a complete dataset, perhaps you aren’t “qualified”, etc… for me The Right It will give you the permission to disregard the generalized expert studies that cost more than you can afford to build your idea, and in a very basic way. It would seem that to master Pretotyping, you must also have an intimate understanding of your customer. If you’re familiar with the work of Amy Jo Kim, you may already be on board with the concept of the “Super Fan”.

Pretotyping will help you discover and understand your “Super Fan.”

As an aside, it seems all too often that companies, individuals, groups, and our ~~leaders~~ politicians get together in a room, or by themselves and decide with the “best research” from “the best consultancy” what’s right for everyone, without ever stopping to ask ANYONE. When a company encounters a boycott, or no one cares, an individual or group’s idea goes nowhere, or the citizens revolt — it is likely because the “best research in the land” was scrubbed of the outliers, sanitized for expensive consumption, and packaged moreso as research to reassure, than research to guide.

The Right It can also be applied in your own life, for it shows you how distill an idea to its essentials and test it. Though the book is primarily for finding the Right It in terms of a product or service, the type of thinking and tactics used can modified and applied to anything that required a commitment of time in the face of uncertainty.

As something of a PS. I said earlier that I thought this was perhaps more groundbreaking now than ever before, that is because in the book AI Superpowers by Kai-Fu Lee, he talks about AI bringing about something of a creative boom — where the future may belong to the people capable of using tools most creatively and effectively, not necessarily those building the tools. (although obviously, from a financial perspective, building the best tools is an epic pursuit). Those capable of Pretotyping, I believe will be those best positioned to fulfill Kai-Fu Lee’s vision.

Graphic Novel Business Pretotype Experiment with Midjourney AI

If you’ve been following Midjourney and other A.I. art tools, you may be wowed by the characters and artwork it creates, and if you’re business minded… you might have thought: “this artwork is AMAZING. But, can it generate re-usable characters on the fly? Enough that I could use it to create an entire book or a graphic novel?”

You have probably seen a few posts on LinkedIn or floating around the internet where someone says “I wrote a book with ChatGPT and illustrated it with Midjourney”. Often, it seems impressive until you see the results. While the story might not be awful, and the artwork is usually stunning — it is often stunning individually, but not in the context of a uniform story. Though Midjourney CRUSHES it when it comes to one off images, a story with a recurring character doesn’t seem doable out of the box. It also seems to do poorly, typically, with crowd scenes. The “continuous” works I’ve seen with Midjourney, so far, either appear to require a dreamlike story to account and excuse blurry nightmare faces and character inconsistencies, OR require excessive seeding which requires regenerating an image over an over until you believe Midjourney can reliably create that say image in a different context.

For me, my thought was, could I do a graphic novel or something sellable with Midjourney out of the box? Something beyond a one off, or “Midjourney Avatar Calendar”.

I didn’t want to spend too much time mastering seeding (it also seemed like it could exhaust credits on the cheapest plan) and I also did not want to deal with Midjourney generating artwork so masterful that if I needed to tinker, would I be totally incapable of producing similar result. Likewise, I did not want to do something so basic that it defeats the purpose of using a tool as powerful as Midjourney.

Therein came some base assumptions on a workaround for Midjourney out of the box:

Use Manga, because most Manga Characters look Similar
Do it in Black and White, because it narrows the options, and two tones are easier to modify in the event that I needed to mess around in Adobe Photoshop.

The biggest thought/concern was how to do a character that looks essentially the same, without using seeding.

I assumed that Midjourney should/would know what a well known celebrity would look like because of the data it is trained on. I also assumed that aside from using only “manga” in the prompt along with a well known celebrity’s name, I might also get better results if I said “in the style of Famous Manga Artist X”. I was very pleasantly surprised. Other things to specify were “length of hair” and also what the character is/was wearing. In my example below, which I think is more extensive in terms of individual frames than other examples I have seen, you will notice — upon examination — that the clothing is not entirely consistent, nor is the car, or the jewelry. In some cases, the character has weapons, in others not. However, in the chaos and action of what is taking place in a given panel, the cutaways and so forth — I think I’ve proven the case that this could be a viable way to do a Graphic Novel.

But would be grateful to know what you think.

I spent less than 8 hours to generate all the images and create the panels in photoshop. It took about 3.5 hours for the panels, because I did that consecutively. I also used ChatGPT and gave brief descriptions of panels and asked it to give me lines of dialogue to use with each description. I didn’t use it all verbatim, but it served as a half-decent starting point to avoid writer’s block. Lastly, as for the story line, I thought of the theft sequence in the movie True Lies and used that as basic inspiration for 5 Act Structure: Act 1 – The Setup, Act 2: The Party, Act 3: The Theft, Act 4: The Escape, Act 5: Resolution. There are a total of approximately 75 images below. Some were spawned simultaneously by Midjourney.

If you’re reading this and know anything about creating Graphic Novels or artwork, I would be grateful to know how far you think this might be from a finished product, and also how long it would take do what is below without AI. Please ping me on LinkedIn to let me know.

Lastly, I used Miyazawa Rie, in the style of Tsutomo Nihei with the –niji 5 flag, along with manga, black and white, ink, and ultra detailed in my prompt. Miyazawa Rie is a famous Japanese actress.

Interestingly (or not), because Manga characters are generally similar looking I wanted to see if I needed to reference Miyazawa Rie, at all, towards the end of the experiment. With that in mind these are characters I got for different celebrities… which, I hope proves beyond a shadow of a doubt that my thesis about using a celebrity icon and a celebrity artist give Midjourney all the information it needs to create characters that are consistent artistically but also not so close as to be recognizable as who they might be.