With every technological disruption, there's a period when early adopters figure out use cases for new tools. There is a hype cycle, a wave of disenchantment, and ultimately a lot of work that goes into normalizing the technology before it's fully understood and accepted.
The design community has been navigating that wave with advances in Gen AI. At Whipsaw, we see a few trends in how design studios are leveraging this evolving tool. At this point in the infant stages of AI, the most value can be seen by applying AI in the early phases of the creative process where the stakes are low, creativity is high, and precision is refined in later phases of product development.
We still believe that designers are ultimately responsible for how they use the tools at their disposal and that the standards are delivering authenticity and value for our clients. Whipsaw’s clients expect excellence and it’s our goal to deliver on that every time. To do that, we ground our approach in curiosity - about the problem, about the tools, and potential solutions.
At Whipsaw we commonly leverage AI tools like Midjourney, Krea, Vizcom, Topaz Labs, and Magnific to occasionally supplement our creation, in addition to language-based models like ChatGPT and Perplexity, and integrated AI features in Notion and Miro.
When it comes to translating product designs from pixels into reality, AI rapidly reaches its limits.
It should go without saying (but it clearly doesn't), that we NEVER input identifiable information into AI tools and that we operate under the assumption that AI platforms are likely scraping data to train models (see: Instagram, Adobe). Our team works hard to use AI tools as thoughtfully as possible, with the highest regard for client confidentiality, given that our work is covered under strict MNDAs.
Some of the most obvious uses of AI are in business enablement activities, including business development. Tools like Perplexity and Chat GPT provide an interesting starting point for summarizing industry trends, identifying market leaders, and synthesizing desk research as background for new opportunities. Agents like Waldo.ai make market research much faster, shortening what used to take days to minutes.
We also leverage AI for initial (anonymous) contract reviews before involving our general counsel. AI can identify certain troublesome provisions and help us ensure that other terms are included, when absent. We always involve experts, but AI helps those of us who aren't lawyers do a better job flagging things up front, which cuts down on fees.
We can also speed up the proposal process by using AI to synthesize several anonymous past proposals into one document for projects that are similar to those from our back catalog. We can use it to identify budget anomalies from historical data and flag project overrun risks.
In proposals themselves, we love to use Midjourney for imagery as an acknowledgment of the starting point for a client's idea. "You want a hovercraft baby stroller?" "Ok!" When clients ask where the images came from, we enjoy sharing that AI is just for setting a general context.
We believe that clever and creative people will use all the tools at their disposal and sometimes our clients are creative too. If they can create something using AI, then we as consultants need to far surpass it. We are confident in our track record, our network of talented partners, and our ability to bring products that consumers value to market; AI can't replicate any of that.
AI is a great tool to speed up research and validation. For example, how can we revise a workshop protocol to speak to an industry better or what's the best way to synthesize themes in 50 expert interviews? Hint: not with billable hours. Furthermore, AI tools can scrape millions of data points from the web that validate or establish research conclusions at a scale that wouldn't be possible or affordable with manual effort. In these ways, AI can be a fantastic support resource for design research activities. While it is not always factually correct, it can provide a starting point for materials that the team can refine.
The real challenge becomes knowing how to best employ a Gen AI agent to best serve your needs. One such area is using the tool to take care of the tedious creation of baseline materials. Instead of crafting things from scratch every time, we can simply use our tuned prompts and inputs to generate content and materials that we then process and format for our needs. These are the things that every designer, strategist, or researcher has to do with any engagement, but it's the busy work that goes on between the important stuff that we would rather be focused on. In a smooth system, we can focus more energy on the original ideas and what we want to achieve with an AI tool, then let it do its thing.
Another use case for image-based AI is in pushing the team beyond where they might be blocked. Sometimes a design will progress to a point, though we believe there's something more beyond where we are. We might add some sketches to an AI tool and ask it to iterate. Rarely will it generate the next right idea, but it will often generate something that sparks the next right idea for the team.
We think AI is good for setting context. When you are designing anything, you need to get in the right headspace about your subject matter, including where the product you are designing will actually be used in real life. For example, if you are designing a product for a specialized vertical market, you need to completely immerse yourself in that world to learn as much as you can about it. You need to go to the site to understand the context of use, talk to users and stakeholders, and become the user yourself. Even when you’re back in the office conceptualizing, it’s ideal to place your concepts in that environment of use, which can be AI images. “AI can act like a visual scaffolding to frame or prop up ideas, but remember, the scaffolding is not the building - it just helps you build it”, said Dan Harden, designer, Whipsaw. AI also helps to set context with colors, textures, and materials.
A caveat here, with AI platforms and other tools like Pinterest, because these tools are trained on large volumes of data and because they are designed to be in some ways reinforcing, (I.e giving people what they want) we are aware that there is a tendency to regress to the mean. All coffee shops begin to look the same because they pull from the same sources of inspiration.
As we develop user interfaces for our digital experience projects, we sometimes use AI to generate early placeholder content to ensure that we are designing with content that directionally aligns with our needs. We also use image generation to create placeholder images that follow the photography style of the final images we are targeting.
For branding engagements, AI agents can serve as a secondary tool for generating broad swaths of content and content guidelines. It is particularly adept at taking in our inputs from naming workshops and contributing to lists of names for the design team to comb through and down select from. We’ve seen lots of overlap with our original ideas, but also have captured new ideas that otherwise would have taken hours to generate through tedious desk research.
As humans, we take all the inputs possible and try to create new and valuable iterations and innovations from those inputs. AI takes all the inputs possible and tries to create the next logical thing. But our work is not limited to creating the next logical thing since we are tasked with creating the "first of" or something innovative. We flat-out avoid being confined by the tools at our disposal. We always seek sources of inspiration that are separate from the tools we use to execute that innovation. This deep commitment to authenticity, which is Whipsaw blood so to speak, is our magic. If a GenAI tool improves that authenticity somewhere in the process that’s ok.
AI does have a place towards the end of the design process as well. When it comes to presenting designs back to clients, sometimes explanatory details and written context are helpful. AI is a great tool to polish written summaries. We might also use AI by asking "What are the likely concerns with XYZ?" or "How might we improve upon this idea?"
Once a final concept has been selected, AI tools like Midjourney and generative tools in Photoshop are extremely useful in rendering CAD in environments. Who among us hasn't used AI tools to fill or remove background elements or to polish imagery? We've been using AI for a long time and now it speeds up the process of placing that conceptual product in different environments to demonstrate its utility.
We are often asked to help innovators convey their vision, long before they have funding to make that vision a reality. Savvy founders understand the efficiency gained by AI tools and appreciate the pragmatic approach to costly work. Great partners will help cash-constrained founders leverage AI tools as they move their ideas through the development process.
AI Does Not…
For everything that AI can do now and might be able to do in the future, there are some serious limitations. Designers are indispensable to our work because of our ability to do the most important things. Designers work with composing systems and workflows with applications across digital and physical experiences. We manage scale and complexity. We make intuitive leaps. Designers understand problems and problem framing in both abstract and pedantic levels; and are adept at shifting the frame when required. Designers are better at organizing information and effectively presenting it in ways that allow stronger decision-making. Designers answer critical questions like “so what…”
In Conclusion, AI + Designers
Great designers are voracious in their search for inspiration and great businesses are efficient in producing results. AI sometimes helps us do both, but it needs to be used at the right times in the process. Early in the process where low risk and low precision are appropriate in creating sacrificial concepts, and late in the process polish and modify for the right contextual alignment. The messy middle requires the experience of designers, who are often determining what NOT to include as much as what to include. Designers are arbiters of taste and translators of human desires, even when those desires remain unarticulated (hello, faster horse).
There is a swath of low-value work that will be expedited (ok, eliminated) by AI, but we believe that gives our design team the space to focus on the thorny problems in a project, which usually involve translating an idea into a human-centric three-dimensional reality. AI is terrible at that.
That said, we are all still learning. If you're using AI in different or creative ways, let us know. If you have questions about how we think about AI, we are all ears. And if you find that ChatGPT doesn't mention Whipsaw in its list of recommended industrial design consultancies (which it does :-), please add your correction, since AI isn't always right. 😉
Share this article
