Generative AI is about to enter its Trough of Disillusionment. We can all feel it coming. All the hype feels like blockchain and the metaverse all over again.
“GenAI can solve everything.” “It will disrupt every industry.”
Lots of exciting words from tech luminaries have been published in the past couple years, but we’re all waiting for the proof. Where is it?
For the last decade, I ran an innovation lab for the North American business at Kimberly-Clark. Kimberly-Clark makes a few products you have probably used including Huggies, Kleenex, Cottonelle, Depend, U by Kotex. One billion households around the world use one or more of their products every day. In the lab, we “played” with new technologies, talked to VCs, and collaborated with universities. All to find ways to grow the businesses.
In the last two years there, I spent most of my time co-leading and co-authoring the Commercial Generative AI strategy for North America, which became the model for the global strategy. But most importantly, I started using GenAI — mostly Microsoft’s 365 Copilot — to build more sophisticated answers to our pressing business challenges.
Those projects included:
• Building creative briefs instantly by pointing to source documents and a creative brief template.
• Pointing to a folder of files, and asking a question that would be built off that information.
• Instantly building content with our brand’s voice for our emails and websites.
• Framing product benefits based on the consumer feedback we got in our surveys.
• Extracting insights, claims, and product concepts in a fraction of the time it typically took us.
And while we sampled many different technologies, we tried everything first in Copilot as a quick, inexpensive way to see what answers we might get. We’d use that to drive our choices on which platforms we really wanted and which features came first, second, and third on our GenAI Product Roadmap.
Three Truths About GenAI
From that, I learned a few truths. One was particularly shocking.
First, GenAI did have the potential to disrupt industries. Second, the way disruption would come was not being talked about in the press. Third, it doesn’t matter whether you pick Copilot or Gemini or Claude; they’re all so powerful, you won’t tap out their limits any time soon. Last, this little thing called “prompt engineering” is the Trojan Horse that will let you build a true competitive advantage.
From that, I have distilled a journey that shows levels of sophistication in your GenAI knowledge and practice. First, if you’re not actually using it first-hand to answer business questions, then you need to — ASAP. As you look at the 101, 201 and 301 levels below, you will see a progression in the understanding of the value of GenAI. That value is directly correlated to how skilled your prompting is, and how well you marry that to the needs of your business.
Let’s be honest with ourselves — we all would like an “easy button” in our corporate roles. Just install [buzzy platform of the month] and your company will be transformed! But to truly get the amazing value that tech leaders are talking about, you MUST put in the hard work to learn advanced prompting. Then, you dig into your company’s core processes that drive your business and use prompting to decrease the time and cost (in some cases to 1%), and increase value similarly.
The Trough of Disillusionment is coming from those stuck in the 101 or even 201 levels, and not pursuing the capabilities of advanced prompting.
The Trough of Disillusionment is coming from those stuck in the 101 or even 201 levels, and not pursuing the capabilities of advanced prompting. It’s not until you have actually seen the potential at the 301 level that you truly understand how dramatic the impact GenAI can be to your industry, your company, your brands.
Here’s what happens at the 101, 201, and 301 levels of GenAI maturity.
GenAI 101 |
|---|
| What You Will Learn: You will learn what it is, what it can do and what it might be able to do, who the major LLM’s are, what you need to watch out for. |
| Prompt Length: 1-3 sentences |
| Most Frequent Uses: You start summarizing documents, writing emails with it and planning your next trip, generating ideas when you’re stuck |
| Attitude Towards GenAI: You’ve found the answers you get to sometimes be amazing, sometimes not amazing. This inconsistency means you’re not sure how much you can trust it. |
| Value Creation: Low – Medium, more personal and individual |
| Technology: Any GenAI LLM |
| Example Prompt: “I’m traveling to Turkey in April with my family of four. I am a journalist, my wife is a pediatrician, my 12yo daughter loves soccer and music and my 8yo son who likes mythology. We have five days in Istanbul and also want to explore a couple places close by for day trips. Can you provide a table of options that describe the location, activities, best timing, approximate costs and how they are suitable to each member of the family?” |
GenAI 201 |
|---|
| What You Will Learn: You will have sucked as much time savings out of GenAI as you think possible. You will have figured out that a more systematic and rigorous treatment of the questions you ask is what drives the quality of the answer. |
| Prompt Length: Paragraph-long prompts with more structure in how you ask questions and how you want the answers structured back to you. You also start finding your answers in a chain-of-thought series of prompts vs. all at once. |
| Most Frequent Uses: Answers that are more usable in presentations. |
| Attitude Towards GenAI: You’re investing more time in prompt engineering. You wonder how the engineer at the GenAI company you work with creates 10 page prompts. |
| Value Creation: Medium – High. You have found specific business use cases that make for good water cooler conversation and get you noticed by colleagues and leaders. |
| Technology: Many of the GenAI startups have a turnkey solution that will do this…or could build it for you in the span of one sprint. Using OpenAI’s API’s, you can also build this solution internally. |
| Example Prompts: 1st Prompt: “I am a strategy leader at a large fashion retailer and tasked with finding new ways to generate revenue. Please build a full competitive strategy of the retail fashion industry including SWOT, 5C’s, Porter’s Five Forces analysis.” 2nd Prompt: “That’s great, thank you! I’m interested in significant trends in society and technology that open new lines of business that include clothing but also add additional service and experience value on top of our traditional clothing. Please provide a table of 12 ideas for new service+retail ideas with the idea, the trends that support that idea, and any relevant supporting data from the SWOT, 5C’s, Porters analysis.” 3rd Prompt. “I work for XYZ retailer. Look at our 10K and that of our nearest competitors as well as the (12) files in [this cloud folder], and provide a prioritization of the 12 ideas based on our right to win. Also provide a rationale for why XYZ should pursue each idea.” |
GenAI 301 |
|---|
| What You Will Learn: You will understand that prompting is just another computer language — only it uses English as the “coding” language. You will know that used with the right context and in the right place, GenAI could disrupt your business and industry. You will understand that GenAI is not the end, but a very powerful means to accomplish what you’re already doing. |
| Prompt Length: 3-8 pages long (if all the chain of command prompts are placed into a single doc) |
| Most Frequent Uses: Very specific use cases that dramatically increase the efficiency of one of your critical business processes that grow your EBITDA. |
| Attitude Towards GenAI: You’re building 3-10 page prompts and re-evaluating your critical processes to find ways to put more robust definitions around how you evaluate the best options/ideas. |
| Value Creation: Very High. You are reducing time and costs by 90%, sometimes 99%. |
| Technology: Option 1: There are turnkey platforms that purport to create concepts but they may not have the right structure and/or rigor that you need. Option 2: You can build your own platform and create your own RAG (retrieval-augmented generation) selection process to pull the files you want referenced. Option 3: You can distill your current process into multi-page prompts that mirror your current business process. Then apply these mega-prompts to the file with your raw interview data to summarize and extract insights and product concepts. |
| Example Prompts: In just the one area of front-end innovation, GenAI can be used for insights extraction, claims creation, and product concept building. Each company has its own definition of the terms in front-end innovation, so you will have to build a multi-page prompt that captures all of it. Once that’s done, you will easily see where you can add more pages to improve the output. Once you’ve built the prompt, you can copy/paste it and edit it to the new situation. |
Next Steps
If you’d like to learn more about the 301 level, join me the next InnoLead members’ meeting, on January 10th, 2025, where I will provide examples that use real, raw consumer interview data (verbatims) from different industries, and apply GenAI to a specific business process that’s critical to many industries — the generation of product concepts. You’ll see firsthand how insights, claims, and product concepts can be created at no/minimal cost and a fraction of the time it currently takes today.
Oh, and by the way: I didn’t use any GenAI technology to create this post. As an innovator, I love the potential this technology brings, but as a storyteller, I believe original content should be… original.
Hugh Park Jedwill is exploring opportunities to continue to build Digital Health Hubs and serves on the Board of Ballet Chicago. Recently, Jedwill was the Managing Director and EXP Labs Leader for Kimberly-Clark North America. Prior to joining Kimberly-Clark, Jedwill founded several startups in the mobile technology space, becoming “Mr. Mobile” in Chicago. He has held key brand leader roles at Procter & Gamble and Palm Inc., and engineering/supply chain roles at Kraft Foods.
