5 Ways We’re Using AI as a Digital Agency Right Now

As headlines about how AI will steal jobs and make humans obsolete pop up everywhere, we decided to concentrate on the ways humans and machines can work together to achieve even better results.

The buzz surrounding artificial intelligence, or AI, has been inescapable (practically deafening) for months now. Beyond the huge deals being made by players like Microsoft and big investments by other MAMAA (Meta, Apple, Microsoft, Amazon, and Alphabet) companies, for business owners, there’s the sudden promise of increased efficiency and faster answers to tough or previously intractable problems. 

At the same time, for workers, headlines about how AI will steal jobs and make humans obsolete have been popping up everywhere, especially in the creative world. When computers can create art or solve complex problems in seconds, where does it leave humans?

We have always been in the camp of embracing new technology, and for us, AI is not different. So we are testing and learning with the many tools powered by AI that have arrived on the scene recently in order to see how they could help improve our processes. We wanted to share some early examples of how they have been able to speed up our work process and allow us to be even more creative humans.

#1 AI as a starting point for code

As developers building websites and apps, we often come across tricky programming problems. We believe that AI can provide a starting point for brainstorming how to tackle a coding challenge.

In the example above, one of our developers wanted to gain an understanding of a new tool we were beginning to work with. Rather than search the documentation and piece together how to use it for our specific use case, we could just ask ChatGPT, an artificial intelligence chatbot, to provide us with an example.

ChatGPT is also able to follow the thread of the conversation and make adjustments based on our input without having to start over. In this example, AI is providing our developer with 1:1 coaching on how to use a new tool, specific to our exact needs. ChatGPT not only gives us the code we need, but explains it, making it a great learning tool.

When our developers can quickly reach understanding about complex coding problems, there is potential to save hours – not just for them, but for clients, freeing up valuable bandwidth for more innovative activities. 

AI can help developers get the knowable, tedious code out of the way quickly so they can focus on the more complex and unique problems. The less time we need to spend recreating the patterns that are already established, the more time we will have to venture into creating more unique, interactive, and usable enhancements.

#2. Using AI to Create Unique Visual and Audio Assets

Artificial intelligence can also do amazing things with imagery, sound, and video. When it comes to web design, one of the biggest needs for a unique and well-designed website is having good visual assets. It’s important to have imagery and video that are high quality, unique to the brand, and not too stock-photo like. 

This means that a company either needs to have an in-house photography team, or the time and money to find, buy, and edit stock images so that they feel unique to the brand. AI has the potential to make that a lot easier by creating unique and royalty-free assets based on a prompt.

AI will never replace a custom art-directed photo shoot or assets made from scratch, but for quick blocking and tackling, or fast concepting, it’s a great option.

For example, the style we developed for the ETR brand and used on most of our insights articles consists of a white line drawing on a grainy blue gradient background. The process of creating this involves finding a stock line drawing that fits the theme of the article and then editing it to be the correct color on the right background.

Could AI do this for us? We tried training an image generating AI, Stable Diffusion, on what this style looked like by feeding it some examples, and then asked it to create its own.

We gave Stable Diffusion examples of our editorial image style – these examples were made by our team in the past.

Here is one of the image Stable Diffusion generated for us based on the training images we gave it.

The style isn’t perfect, but with more fine-tuning it could get pretty close. We modified one of the line drawings it gave us to make it high resolution and have the right background color, and used it for the cover image of the article.

We also tried coming up with some unique assets with DALL-E 2 to help convey a concept in a presentation – someone talking loudly on their cell phone during a boat ride. Based on this, AI image generation still has a little bit of fine-tuning needed to get out of the uncanny valley of not-quite-right imagery. The results here are a bit funny, but fine-tuning prompts can result in some amazing imagery.

Prompt to DALL-E “illustration of a person holding a cell phone with loud sound coming out of it with sound waves on a tour boat. Others are annoyed at the noise”

Companies like Adobe Stock are embracing generative AI content, allowing contributors to submit and get paid for the content as long as they have the rights to it, though who exactly should have the rights to generative A.I. art is still hotly debated.

#3 Using AI for quantitative user research

When working with quantitative user research data, or data that is number-based and includes larger quantities of information, we often have to open a spreadsheet and start number crunching to analyze our results. AI has been helping our research team remember and apply formulas we need to use in context.

Again ChatGPT is a great learning tool for determining the right function to use and giving instructions for how to use it in context.

At this point, I was just using ChatGPT as an alternative to Google. I knew this was a simple function, but I just couldn’t remember what it was. Rather than dig through documentation or Stack Overflow posts, I can get my answer directly from ChatGPT.

#4 Using AI for qualitative user research

Qualitative user research is often considered more “human-focused” and all about empathy, so it seems odd at first to employ AI for this method. However, just like our other techniques, we only seek to use AI in this way to speed up our process where we can and stretch our creative thinking. 

Part of preparing for qualitative user research like interviews involves preparing a list of questions to guide the conversation. Could AI help us with that?

Now, this isn’t the exact question list we would go straight into an interview with. But prompting ChatGPT to come up with questions for us based on different scenarios can help us brainstorm what questions to add. Maybe asking about privacy and security related to the topic wasn’t something you had thought of yet. Thanks AI!

What about AI teaching us empathy? It feels like a stretch, but ChatGPT was trained on a massive amount of data that probably included written accounts of the wants, needs, goals, and frustrations of people from all walks of life. Could AI help us get into their mindsets?

Data like this can help us get into the mindset of our target audience prior to interviewing them, or help us write helpful website copy later on. Of course, we should always validate that the information we learn about an audience from AI is accurate, and this is no substitute for human-to-human interviews, but we think it can provide a really helpful start.

What about after the interviews have been conducted and it’s time for analysis? This is where we think AI can really speed things up. After we have a boatload of interview transcripts, the next step is to pore over the data, looking for themes. It’s very time-consuming. AI could easily scan these transcripts in seconds and bring useful patterns and themes to our attention. Again it’s no substitute for the human mind, but it could act as a second pair of eyes, identifying areas we may have missed. 

We fed ChatGPT a small sample from a recent interview with a veterinarian to see what it could come up with. (Interviewee name changed for privacy)

ChatGPT was able to succinctly summarize the core points made in this interview. 

Services like Notably.AI, which claims to help you analyze user research faster with artificial intelligence are popping up to address this opportunity.

#5 Using AI to generate ideas and create new products

Continuing with the theme of summarizing information, ChatGPT is really good at taking in all sorts of information and providing a succinct summary, explanation, or edit. This has endless uses for business of all kinds, like summarizing a lengthy email exchange.

AI can save us all a lot of time if it can cut through the jargon and formality of email etiquette and just get us the highlights of what needs to get done.

Many AI technologies are giving access to their APIs (Application Programming Interfaces) to enable developers to create apps using their systems. Our Director of Engineering Ryan Ogden has been experimenting with this possibility in various ways, such as a plugin that can be added directly to email applications for quick summarization.

Soon Gmail itself should be rolling out these features. Development of AI products is moving at a breakneck pace, and everyone is trying to hop on the bandwagon.

It was recently Ryan’s birthday, and clearly his enthusiasm for AI has spread throughout the office, since this is how coworkers chose to celebrate:

An explosion of AI-powered apps and debate

We predict that the trend of artificial intelligence isn’t going away anytime soon, since it has already proven extremely useful in these early days of its widespread availability. More and more apps will begin integrating A.I. as a part of their offering, and we can definitely expect much more excitement and debate to come.  

Don’t expect services like ChatGPT to continue to be free forever. The app currently states at the bottom of its page that it is in a “Free Research Preview” as the team collects data on its usage and continues to fine-tune the technology. GPT plus was recently announced for $20/month, giving you access to the latest updates and faster response times.

The ethics of AI will surely continue to be a major topic of discussion. It’s our duty as early adopters of this technology to think about the impact it will have on the world, remain diligent on the ways it could potentially be abused, and use our human ability to think critically to evaluate whether it is providing accurate and useful information.

Bonus use case: AI for writing conclusion paragraph

It’s only fair to end an article about all the great uses and shortcuts we can gain from artificial intelligence by having artificial intelligence wrap this up for us. Take it away, ChatGPT.

In conclusion, as web design agencies face the challenge of staying competitive in a rapidly evolving digital world, they are increasingly looking towards artificial intelligence (AI) to optimize and streamline their processes. The use of AI in web design is still in its early stages, but it has the potential to revolutionize the way we approach web design. From assisting developers with coding and user research to generating unique, on-brand assets, AI can help improve productivity and efficiency while freeing up creative time for designers to focus on more complex, higher-level tasks. While concerns around the impact of A.I. on the job market continue, it is important for web design agencies to explore the potential of these tools and embrace them where appropriate.

This article was originally published by ExpandTheRoom, a company acquired by Infinum in 2023.