Let’s talk about artificial intelligence (AI). I don’t know if you’ve heard (you have), but it’s a pretty big deal. Many years ago, we began using AI and machine learning (ML) tools to enhance our marketing services. Early last year, we built a custom solution to drastically reduce the price of online language translation through the use of deep learning, while crafting the best integrated translation experience on the internet.
This year, we’ve made a decision to invest heavily in AI in three ways:
- Bringing ChatGPT Plus to every team member as a personal AI assistant
- Building additional AI solutions within our preferred ecosystem
- Investing time and money in AI to enhance our internal operations and bring new services to our clients
While these decisions are exciting, and we have immediately seen the positive impact of our investments in AI, this new world is understandably intimidating and frightening to many. We value transparency and communication at Mostly Serious, and in that spirit, I’d like to share more information about our history leveraging AI and ML, an overview of how we are currently utilizing ChatGPT on our teams, a brief look at the near-term future, and share our intentions with AI moving forward.
Our Experience in AI
AI in Marketing & Automation
The recent explosion of new large language model (LLM) tools like ChatGPT has made many people forget that AI and ML tools have been prevalent in sales and marketing for many years. Our team has harnessed tools to automate and optimize marketing processes, such as lead scoring and predictive analytics, since 2018, using services like HubSpot Marketing Hub and Salesforce's Einstein AI technology. These tools are similar to the recent surge of AI solutions in some ways but highly targeted toward sales and marketing operations and not trained on nearly as much data as OpenAI's GPT-3 and GPT-4.
The primary benefits of using these early AI tools in sales and marketing were improved lead generation, better personalization tools to appeal to website visitors, increased efficiency and time savings, and the ability for our clients to more easily use data to make better decisions.
Deep Learning for Translation Services
In early 2022, we started working with a cutting-edge language translation tool that uses deep neural networks to produce translations that are more natural-sounding and accurate. After years of seeking a solution that improved upon existing translation tools, like Google Translate, and could save our clients tens to hundreds of thousands of dollars in manual translation costs, we found DeepL.
Manual translation services are expensive. There's a great reason for that as it's important for any brand to meet their customers where they are, and speaking their language is a prerequisite. If your customers can tell that you didn't put the time into making their experience as great as others, your brand takes an immediate hit. We understand, though, that if we can automate translation services using AI and achieve the same outcome as manual translations, our clients can dedicate those funds—in many cases, six figures—to improving the customer experience in other ways, which is the best outcome for everyone involved.
Our favorite way to explain the power of DeepL is through a client story. One of our clients evaluating translation services is a native Spanish speaker. She wrote an email to her husband in Spanish, then had DeepL translate it into English. When she got home, she asked if he found anything odd about her email, and he said he didn’t. Other translation services didn’t come close in her view.
Craft CMS and DeepL: A Powerful Translation Solution
We leverage Craft CMS for the majority of our custom website builds, largely because the authoring experience is the best available. While our agency isn't exclusive to Craft CMS, we have yet to find a client that chooses an alternative system after seeing the benefits Craft provides. It's simply great in every way.
Because the authoring experience is so important to us, it was essential that we found a novel and seamless way to use DeepL with Craft CMS. That's why we decided to build a custom solution that allows our clients to translate content with a single click of a button at any level—from a single block on a page, to an entire page, and even the entire website. The result is the best integrated translation experience on the internet.
Looking Ahead: Upcoming AI InitiativesBeyond our work with marketing intelligence tools and our custom DeepL + Craft CMS solution, in 2022 we launched a focused initiative to bring new AI solutions to our clients. Our mission with our next release is to create the best AI-backed web content authoring experience in the world, directly within Craft CMS. We're currently in closed beta and will be launching in May 2023.
Balancing AI with Human Creativity
We believe that AI advancements have opened the door to great new opportunities. Increased efficiency is only one benefit among many, and companies that over-leverage on that benefit will be left with irreparable damage to their brand. Through upcoming innovations, like Google's radical search changes, customers will have more control over how they find the products, information, and resources they need. If they land on your website, and you're overwhelming them with generic AI-generated content that has no soul, they're going to leave.
Now, more than ever, companies must dedicate resources to building a strong brand and an engaging, personalized customer experience. Not only will an investment in the holistic customer experience, from first encounter through customer advocacy, distance companies from their competitors, it will also allow them to charge a premium for their products and services.
Riding the AI Wave Responsibly
A long-time principle at Mostly Serious is not to chase the new shiny object. This is especially difficult for me, an avid lover of shiny objects. We follow trends closely and understand where the world is going, but that doesn't mean we will subject our clients to the whims of the latest fads or unproven technologies. Instead, we prioritize stability, reliability, and proven solutions that best serve our clients' needs and goals. For example, while NFTs (non-fungible tokens) offer something interesting in specific situations, we didn't buy one of those monkeys.
AI is different. Yes, there are certainly fads within the AI space, like a current obsession over AutoGPT, a tool that, in its current iteration, never seems to actually arrive at a useful outcome. (As a lover of shiny objects, believe me, I tried). But the trend that is AI will have society-shaping impacts. Our stance on AI is that we can either choose to ride the wave, or risk being crushed by it. We feel our clients are in a similar position, and it's our responsibility to ensure they benefit from stable, reliable, and proven solutions and do not get caught up in the latest artificial shiny object.
Bringing AI to Teams
Intentionality is at the heart of good organizational decisions. While I was writing our Responsible AI Usage policy at Mostly Serious and working with our legal team to ensure we were making decisions that were safe for our organization and right for our clients, I talked with a number of leaders at other agencies, manufacturing companies, and healthcare institutions. Consistently, I heard that people within those companies and organizations were using AI, but that there had been no instruction by the leaders of the companies, no policies and procedures created, and no directive given to staff on how AI should be used.
Our introduction of ChatGPT into the organization was quite robust and included adding new policies, providing clear instructions on how to use the tool effectively and responsibly, ongoing training, and an internal Slack channel to share our experiences and learnings. However, the investment of time has already generated positive returns, as our team finds creative and innovative ways to use our tools to better our processes and outcomes. We believe that AI offers our teams a competitive advantage while allowing us to better serve our clients. Early experiments have shown a 30–50% increase in efficiency from people using ChatGPT without any training or instruction. We hope to exceed those outcomes with an intentional approach, allowing our team to focus more of their time on strategic, creative, and novel work.
We rely on our values, principles, and demand for high-quality work to drive our decisions for how to employ AI.
Lastly, we rely on our values, principles, and demand for high-quality work to drive our decisions for how to employ AI. Put simply, AI isn't capable of creating work on par with our internal standards. However, it's a fantastic tool to get past the blank page problem, or to help solve a tricky development bug, or to brainstorm ideas. I used GPT-4 to assist in the structuring and topics for this blog post, but it didn't write a single published word. I wouldn't be living true to our values if I allowed AI to talk for me on such an important topic.
A final note for leaders who aren't ready to introduce AI tools to their teams: In times of extreme change and volatility, communication is key. Even if the communication conveys that the organization is choosing to wait and see, which is entirely understandable.
What Comes Next?
I cannot pretend to know what the future of AI will bring. I don't believe anyone can. What I can do, and what we will do at Mostly Serious, is learn, grow, and evolve together as we individually and collectively discover new ways of working.
Most importantly, we'll be human.