How Sho.ai Uses Artificial Intelligence To Help Manage Brands
by Rudy Sanchez on 09/14/2023 | 7 Minute Read
Artificial Intelligence (AI) has been a hot, contentious topic of discussion in the design industry as more generative products have become available to the public, like ChatGPT, Midjourney, and DALL-E. Conversations about AI's impact are occurring in nearly every industry and sector. Some are looking for ways to incorporate generative AI into their organization to optimize and improve operations, while others fret over the possible consequences of this technology.
Sho Rust, founder of Sho.ai, wants to improve how organizations manage their branding and create new branded content through AI.
"Sho.ai is an AI platform for premium content creation. Typically, that means brands getting out really exciting stories they want to tell," says Sho. "Usually, they have a beautiful story to tell, like why they founded the company and how they came up with products or services. But then they have a hard time distributing that story across different platforms, audiences, and languages.”
“That's where AI can really help,” he adds. “Whether creating visual assets or text or in different languages, everything is based upon an AI trained on who you are as a brand.” What Sho.ai delivers isn’t the typical AI-generated content anyone can whip up on ChatGPT or Dall-E. Instead, it trains on your own brand data, whether it's brand guidelines, assets, documents, copy, or posts, to tailor generated content about the organization. Once the AI knows everything about your company, it can behave like the brand and create brand-specific material, whether it's sales emails, visual assets, graphics, or anything that can help grow your story.
The more about the brand fed into Sho.ai's data engine, the better, as the platform can take on more complex projects. Rust says Sho.ai understands the brands it's trained on so well that when users begin a project with a prompt, the AI begins probing with questions on the intent of the messaging.
"It's like having someone on your team to talk to, someone who knows everything about your business, customers, benefits, how you do things, or everything you've uploaded into a platform," Sho says. "That means instead of doing any of this prompt engineering or hacking in all of the brand context and everything the AI needs to know within the request, you simply ask things like, 'Hey, write the sales email for me,' or, 'Let's run a campaign in Times Square.' Then Sho.ai already knows that this campaign is for this company or this particular product they're probably working on."
When asking a lot of generative AI software to create something specific, like, say, copy or art that fits within a brand guideline, users have to provide background or go through so many revisions that any benefit to using AI is negated by the amount of work and time required at the prompt. Sho.ai, being more familiar with the brand and its consumers, doesn't need the extra direction to get the right messaging and tone, use the right colors, or correct typefaces.
"With AI right now, what people do is have the AI read a first draft, and then tell it, 'Hey, be wittier,' or, 'When he talked about this, Connor mentioned these different things,' and this ends up taking more time than it would have taken just to write the email in the first place," Rust explains. "In our platform, a customer question, for example, can be inputted, and the AI will give a perfect answer back, better than the human. The only way you can achieve this is by having a data engine, which our platform provides, to give the AI all the context it requires. We focus on ensuring that humans have the tools to affect the core memories of the AI better than any other platform. Every other company focuses on building foundation models or creating interfaces for these foundation models. We spent the last six years building a platform that trains these AI models to act consistently, follow the brand values, and mitigate some of the risks and problems with AI."
Though showing promise in helping brands create consistent content quickly and very much on-brand, Sho has a more challenging time explaining how generative AI isn't coming for your design job. Rust believes that AI products like his serve as an equalizer for companies smaller than Google or Apple that don't have the resources to build their own AI tools.
Moreover, Rust says that now is the time to familiarize yourself with these tools.
"AI has a lot of promise, and people should be somewhat concerned about not adapting to this technology, if only for the increased productivity and efficiency," Rust says. "There are some things that AI is extremely good at," Sho said. "AI applies to almost every job in the market, so people should adapt." Sho also explains that now his program is being used by students at ArtCenter College of Design, and according to him, it's helping students not only create better work, but it levels the playing field a little bit for students whose second language is English. Even if those students don't understand a syllabus in English, with Sho's platform, they can talk to the syllabus in their native tongue without needing a supercomputer at home. What's more, instead of needing pricey 3D software, every student can visually explain their ideas and iterate with relative ease.
Besides students embracing Sho's platform, the CEO of The Futur, designer Chris Do, has trained Sho.ai to create a "Do Bot," feeding the data engine seemingly countless hours and pages of his content. The result is an AI bot that responds to prompts just as Do would, rather than offering generic guidance based on the consensus of many people’s ideas that the AI trained on. For example, if someone were to ask Do Bot for advice, the AI starts asking questions to understand intent and then responds as if it were Do, mimicking language, tone, and content.
Perhaps it's even too impressive. It's one thing to consent to being used as a source of training for an AI bot that is supposed to think and create like me. But what if an employer or competitor wants an AI version of myself to help create new content that I didn't consent to? Is there a scenario where my employer no longer wants to retain my services and settles for an AI version of myself that can train using every piece I’ve written using tech like Sho.ai?
"It's definitely happening right now," Rust says.
"I can go to any AI system and say, 'Act like Steve Jobs, and your role is to behave like Steve Jobs; I'm gonna ask you a bunch of questions and respond as if you were a Steve Jobs,’” he further explains. “And then it'll say, 'Hey, I'm not Steve Jobs, but I'll play this role for you.' I see a huge issue when the AI goes online or starts misrepresenting itself as someone notable, especially if they're alive. That creates true confusion in the marketplace. A lot of laws immediately start to come into play. It's one thing to look at the Mona Lisa and be inspired to create a sketch. But if I went to the AI and said, 'Create a piece of work like Banksy,' and I signed it as Banksy, and I sold it online as Banksy, I'm breaking a lot of laws, and I go to jail. I think there are laws to protect against this and protect individuals from certain data."
"Let's say they try to credit Rudy AI, and your company used all your previous writing and different things like that," Sho begins. "There are still certain insights not in beta that you can add to the system that would be able to produce better writing. Or there are these different insights, or other mechanisms, and the cognitive piece that needs to be added to the AI. Now, the real advantage is that Rudy combined with the AI Rudy writes more and identifies things it doesn't have to worry about, like writing ledes."
I suppose there is some truth to Sho's assertion that it's already happening and that some laws are in place to discourage the misuse of AI. Ultimately, "go learn how to use AI to be more efficient" isn't exactly reassuring or good at dispelling apprehensions of the long-term ramifications of technology we still don't fully understand as we're still very much in the early stages of how these tools will be deployed.
Creative work generated AI that's “good enough” for some applications and use cases already exist and will only get refined and improved over time. The inevitability of AI automating even creative jobs might bolster the notion that creatives should learn to incorporate this technology into their work.
Adapting to an inescapable future with AI is certainly a pragmatic outlook. Still, it doesn’t address or even begin to describe the possible implications of having machines be able to mimic humanity to the point of fooling its creator. Perhaps the legal system will suffice in mitigating any negative consequences, as Sho says, or we reach a point where AI turns the tables on us, and whoops, we started the Age of Machines.