AI-First Blueprint: How to Build a Business Around Custom GPTs

Full video transcript

All right y’all, tuning in here for the final keynote session of the day, joined with John Munsell, going to be talking about building an AI-first culture, a leader’s guide to scaling AI across your organization. Of course this is the final session of our day one of three of our live Scale with AI Summit, made possible by our sponsors Howdy and Gamma, and I’m excited to be joined today with John, who — just looking through your outline of what you can be presenting, John, there’s no one better to speak on the subject than you, I believe, and I’m really looking forward to your session as we wrap up today’s live. How are you doing, John? I’m doing great, I appreciate it, although the outline has changed dramatically since — I hope it’s still on the same subject.

Anyway, I just needed to drive home the point a lot more, and I’m also super glad that you use Descript so you can take all the “ums” and “ohs” out of this when it’s finally recorded. Absolutely — all right, thanks again Darby and welcome everybody. I know this has been probably a long day if you’ve been with us for seven or eight hours. I want to make sure that you know from the start that this isn’t just another presentation on why AI matters — this is really, in my mind, more about survival. And in the next hour, what I’m going to share with you is a solution to a problem that I don’t even think most businesses know they have, and it’s one that could literally make or break your company’s future. More importantly, I’m going to show you exactly how to make sure that you’re on the winning side of this problem, and it’s not going to be a pitchfest — although at the end I will make an offer to those who want to go deeper than what I’ve given you. But I’m going to give you the exact tools, I’m going to give you the frameworks, I’m going to give you the ability to download those after the fact.

So I want you to think about this: everybody’s vision is that AI is going to speed up everything you could possibly imagine, and it’ll either free up people to focus on other things, or increase efficiency and profitability. But what if those successes actually become your biggest threat? Imagine this scenario: your team implements an AI initiative and productivity goes through the roof — you’re crushing it, and now instead of something taking 40 hours it takes four. So now you’re left with a choice: do you give people time off, do you lay people off? You don’t want to lay people off, they’ve been with you so long — so you decide to reap the benefit of those gains and absorb the new profits into the corporation. It’s only a matter of time before your competitors learn how to do the same thing, and they start playing with pricing. Now you’ve got price cuts, so you’re sitting there thinking, “we just gave people paid time off, now what do we do — unpaid time off?” But your competitor cuts prices again because they’ve also gained efficiency, and now you’ve got real margin squeezes, and at that point you start laying people off. Then everybody else starts laying people off too, and that’s the big fear right now: a reduction in the economy because people are no longer earning a salary that goes back into the economy, which means reduced consumer spending, which creates a market contraction.

That’s the nightmare scenario everyone is so concerned about with AI, and that’s why I say we’re really at the beginning of what I’d call the AI arms race. If you sit around with a wait-and-see attitude, you’re going to lose. So my job today is to get you thinking not just about how you can compete, but how you can enter the race, stay in the race, and be one of the businesses that survives and thrives.

So who is this guy? My name is John Munsell. I am the CEO of a company called Bizzuka — we’ve been around for almost 25 years. We started as a software development company building web applications, built our own web content management system, and gradually grew into mobile app development using machine learning and other tools. We then became more of a digital marketing agency, which I sold off about two and a half years ago because I saw the new frontier was AI. We’ve been working with AI deeply for a little over four years now, and I’ve personally put in at least 8,000 hours mastering the tools — I’m not kidding, I’ve been working 14-hour days, seven days a week, for at least the last three and a half years. It becomes an addiction, I might add. I also created a tool called the AI Strategy Canvas, which I’ll walk you through in a bit. I’m also the author of a book called Ingrain, coming out in December — I’ll show you how to get on the waiting list — and I’m an adjunct instructor of AI at LSU.

Like I mentioned, this is the scenario that keeps CEOs awake at night, but if you want to survive and thrive you need to act quickly and start creating an AI-first culture — getting everyone in your organization thinking about how to go AI-first, how to simplify, ideate, and create using tools like Descript and Gamma, among many others. The more your organization thinks that way, the more you’re able to thrive as this wave takes off — and it’s already taking off at record speed.

There are really four critical mistakes most companies are making right now. First, fragmented AI implementation: somebody in marketing is using it, somebody in HR is dabbling with it, somebody in IT is dabbling with it — everybody operating in silos, none of them in sync. Second, no systemized training, so everyone is self-taught, which leads to chaos because nobody is doing it the same way. Third, no scalability — no process anyone can share or grow with. Fourth, no AI leadership, which is a critical function for making all of this gel. The warning signs: departments using AI differently and reinventing processes almost daily; training that doesn’t stick because nobody is doing real training (people are going to cheap online courses or just goofing around on YouTube); scattered AI tools, with three or four tools doing the same thing and nobody standardizing the process, leading to frustration and no one knowing who to go to for help. Meanwhile, competitors that have common understanding, scalable processes, an ongoing training system, integrated tools, and dedicated AI leadership are experiencing accelerated growth.

77% of companies list AI as their top priority, but 87% of them struggle to find properly trained talent. If you can’t find talent, you have to breed it from within — which is roughly seven times cheaper than replacing an employee with someone who already knows AI. You’re not going to be replaced by AI, you’ll be replaced by somebody who knows AI. As an employer, the last thing you want to do is let someone go, hire someone new, and spend a year getting them up to speed on your culture, processes, and SOPs, when you could upskill a loyal, productive employee in two to four weeks and have them insanely productive.

While most companies recognize AI’s importance, they’re failing at implementation. There are three pillars of AI fluency we want to build in an organization. Pillar one is AI literacy — a shared language, so everybody speaks the same language when it comes to AI. Pillar two is a scalable prompting framework — a way to literally hot-swap variables so someone else can use the exact same prompt structure for a different purpose. Pillar three is unified AI leadership.

On pillar one: when everybody speaks the same language, adoption, efficiency, and innovation truly accelerate. Gartner’s AI hype cycle shows a peak of inflated expectations followed by a “trough of disillusionment,” and generative AI is currently on that downward slope — meaning people are realizing AI is a lot harder to fully harness than they thought, given biases, hallucinations, integration challenges, processing power limitations, and security and safety concerns. Scaling AI in an organization starts with AI literacy, and the key to scaling literacy is that everybody speaks the same AI language at every level of discussion: strategy development, cross-functional initiative planning, and prompt engineering. That’s exactly why the AI Strategy Canvas was created — a free downloadable tool, structured like the business model canvas, divided into two quadrants (value-creating prompts on one side, governance/restriction on the other).

The canvas is used at each of the three discussion levels. At the strategy level, an initiative is first broken into one of three types: is it about innovation, customer engagement, or operational efficiency? Once you decide the focus, you use the canvas to further define the initiative through a series of blocks: who are the beneficiaries (target audience); what does the AI need to know about the company; what products or services are impacted; what other context is needed; what role is AI being hired to play; what internal or external resources are needed; what rules must be followed (including relevant legal requirements); and how will success and progress be monitored. At the cross-functional planning level, the same canvas is used with a team assembled from the relevant departments (customer service, HR, IT, legal, etc.) to deconstruct the initiative again through the innovation, customer-engagement, and operational-efficiency lenses, refining the audience, aligning departmental views with company goals, and building a detailed execution plan.

Munsell shared an example from a client, John Thompson, whose project-management software company (Exapro) used the canvas to move from feeling lost and confused about prompting to having a structured, repeatable process — eventually building the canvas into the company’s own software so it could isolate a reference database of frequently-asked customer questions, rewrite them, and answer product-specific queries while refusing off-topic questions (their test case: asking it “who is Beyoncé” and having it decline to answer).

Pillar two, scalable prompt engineering: when everyone uses the same prompt structure, efficiency and effectiveness increase and success multiplies across the organization. Traditional prompting has no shared structure — everyone has their own methodology. A scalable approach gives complete strategic alignment, consistency across departments, efficiency in AI interactions, and faster, better results — and everything created is reusable and adaptable, so the learning curve accelerates across the organization. Using the AI Strategy Canvas, a scalable prompt is built from stacked containers: target audience/persona (niche, fears, goals, past failed attempts); company information; product/service information; context (the user’s own thoughts, not just a bare request); role (e.g., copywriter, SEO expert); style calibration using numeric parameters (e.g., “humor = 4 out of 10”) to avoid “prompt conflict” between a role’s inherent writing style and the desired voice; resources (internal links or external URLs); rules (words and phrases to avoid, such as “I hope this email finds you well” or “skyrockets”); and finally the request itself, placed last so the prior context reduces conflict and produces a better output. These containers form a “prompt stack” that others can reuse by swapping just a few variables — for example, swapping the persona to write to a completely different audience — without rebuilding the prompt from scratch.

More audience testimonials followed: John Thompson described using a shared prompt to write a co-marketing article about an unfamiliar HR product by combining a URL analysis with his own product knowledge, producing a piece the partner company called “fantastic.” Tracy, writing a book on prompt engineering for law professors and students, described the approach as “revelatory” compared to what she calls “word vomit” paragraph prompts, which produce inconsistent results, force people to reinvent the wheel, and keep expertise siloed. Fabio, an LSU professor applying for a multi-university research grant, said a custom GPT built with this process gave him far more specific results than before, and drew praise from a grant reviewer at another university for how well the writing represented the linkages between science and risk-management communication. Jeff, an independent fractional CMO, said building a client’s custom GPT with the canvas took five hours instead of the one to two weeks it previously took, and that the output quality “went from grade five to grade twelve in one go” once he defined a clear structure with good and bad examples.

Pillar three, unified AI leadership: leaders must oversee the process, bring everyone together, and encourage an environment of experimenting, learning, and sharing so AI-first thinking becomes part of the company’s DNA. This unified leadership sparks creative solutions, accelerates companywide adoption, builds enthusiasm and trust, drives responsible innovation, and creates collaborative success — a flywheel that improves over time. Without central leadership, nobody knows who maintains standards, shares successes, drives adoption, or ensures consistency. With centralized leadership, there’s clear direction, known resources, and shared victories. Two common approaches: an AI Council (a cross-functional team that meets regularly, best for large organizations just getting started — Munsell has seen this often in universities) or a chief AI officer (full-time or fractional), who becomes the organization’s single point of leadership, clear line of accountability, and best suited for rapid transformation. Unified AI leadership functions as the organization’s “AI success center” — ensuring frameworks like the AI Strategy Canvas and scalable prompt engineering are properly implemented, serving as the go-to resource, curator of best practices, and amplifier of successes (and of lessons learned from mistakes), so that when someone discovers a better way to use AI, that knowledge spreads throughout the organization via a hub-and-spoke system.

To sum up the blueprint for building an AI-first culture: build AI literacy until it becomes AI fluency; build a process for scalable prompting rather than letting people prompt without structure; and put someone in charge through unified AI leadership, since AI collaboration and innovation depend on someone overseeing the whole system. Munsell argues that neither IT nor marketing is the natural owner, since each brings different influences to the process — which is why he favors a chief AI officer, full-time or fractional, someone who understands AI at a strategic level and can rally the organization. The goal is a mindset where “if they can think it, we can build it, and then we can scale it.” The two paths forward: stick with fragmented AI adoption (scattered efforts, wasted resources, shrinking margins, widening competitive gap) or commit to an AI-first culture (accelerated growth, rapid team adoption, a unified approach, and a clear strategic framework to stay competitive).

Munsell closed with two offers: his upcoming book, Ingrain (targeting a beginning-of-December release, with a waiting list at bizzuka.com), and his “AI Mastery for Business Leaders” course, which teaches the AI Strategy Canvas and scalable prompt engineering in depth, includes a Notion database of prompts, requires a capstone project (a custom GPT specific to the participant’s own job or business), and includes six months of community membership with weekly live office hours. The live-taught version of the course was originally $3,600; the self-paced, community-supported version is normally $1,479, and Munsell offered summit attendees access for $739 using the coupon code “scaleai,” crediting Darby’s own prompting techniques as the inspiration for the scalable prompting system.

In the Q&A that followed, Munsell said the course applies broadly — from solo consultants to large corporations (citing ten LSU professors, marketing professionals, and independent fractional CMOs who have gone through it) — and mentioned a second, forthcoming certified-trainer track planned for December or the first quarter of the following year, alongside an LSU certificate participants can add to LinkedIn. Asked whether the course would suit a special education advocate’s work, Munsell said he couldn’t be certain without understanding her specific audience and workflow in more depth, but pointed to a money-back guarantee and said he struggles to think of any profession — short of physical labor — that couldn’t benefit from applied AI skills.