Meet Our Newest Teammate, AL

Why we built and deployed our own AI system, powered by behavioral science.

Insights from our Technology and Innovation Director, Elijah Kleinsmith

As illustrated in The Little Prince, human creativity and vision can transform the mundane into the extraordinary. It’s about seeing potential and beauty where others might only see simplicity. When we use our imagination, we can reframe and elevate the ordinary, turning it into something meaningful and significant. Curiosity, driving this transformation, encourages us to explore these new perspectives.

Consider the octopus: it is older than the dinosaurs, yet exhibits remarkable intelligence and complex behaviors such as using tools, solving puzzles and deploying sophisticated camouflage techniques to avoid predators. Though our closest evolutionary relative is a worm-like organism, we find the octopus to be a curious creature, a lot like us humans. Intelligence comes in many forms.

At Signal Theory, our focus is more on humans. Every day, we wake up excited and ready to tackle consumer behavior challenges that test our intelligence, intuition and creativity. Going forward, the ways in which we augment and enhance those qualities may never be the same. And it is curiosity which drove us here.

Introducing AL

Earlier this month, we announced AL, our custom AI system which fuses advertising and behavioral science. The name “AL” is a tribute to one of Signal Theory’s original founders, Al Higdon. Perhaps one also sees a wink and nod to 2001: A Space Odyssey. AL was built for humans by humans.

It was very important to me personally that we make the core features of AL available to all Signal Theory employees. By building our own system, we are able to remove limitations associated with having a limited number of user licenses. By opening up an easy, private way for everyone to experience the power of large language models (LLMs), we enable diverse contributions to our collective knowledge of how AI can be leveraged at the firm and for our clients.

AL integrates into existing workflows to augment and enhance our capabilities. This effort toward introducing AI to our advertising workflow was and continues to be developed through close inspection of our processes in order to align it to our unique approach. AL is trained on behavioral science research, aiding us in our human-centered purpose of connecting people with joy, comfort and meaning in an ever-changing world. And what an ever-changing world it is.

Capabilities

AI may be the newest buzzword in town, but AL is just a work buddy. To make the benefits real, we knew we had to ensure the system’s security. Since addressing those concerns from both a technical and legal perspective, we have been hard at work adding to AL’s capabilities nearly every week.

Here are just a few of AL’s proven capabilities:

Research assistant returning cited summaries from high-quality sources
First draft local store marketing plan generation based on first-party data
Agronomist field notes analysis and multi-modal processing
Consumer survey and quiz analysis, reporting and response
Enhanced marketing personalization and data enhancement
Automation of routine tasks, including clerical tasks like logging PTO
Moodboard generation and creative brainstorming assistance
Website traffic, social listening and engagement analysis and reporting
Thought leadership and proactive idea generation
Intelligent and fully customizable voice assistance

By building AL securely from the ground up, we have set ourselves up to iteratively improve our system moving forward. New features are rolling out all the time, further integrating AL into our workflow and freeing up time to contribute in increasingly meaningful ways. I often joke that AL sends more emails than I do. That’s becoming increasingly true as we enhance AL’s capabilities and enable more automations. It is exciting to consider what possibilities this opens up for us heading into the future.

“AL helps free up time for the expert humans behind the work to operate at higher levels, providing deeper, more meaningful value.”

Of course, none of these applications would be possible without the human input that makes use of this new extraordinary technology. As Chief AI Scientist at Meta, Yann LeCun, writes, “Agency (and planning) can’t be a wart on top of Auto-Regressive LLMs. It must be an intrinsic property of the architecture.” While AL is able to assist with a wide range of tasks, it lacks intention, planning and agency. By helping Signal Theory employees with launching points and augmentation on certain tasks, AL helps free up time for the expert humans behind the work to operate at higher levels, providing deeper, more meaningful value.

Adoption

AL has multiple interfaces, including a chat interface that most are familiar with. AL also has its own company email address which enables it to do a lot of real work like regular website analytics, media coverage alerts or on-the-fly social listening and research reports on any subject. There are even a few easter eggs like the number in AL’s email signature – if you call it, you’re connected with a voice interface where you can actually have a verbal conversation with AL.

I have gotten some really encouraging anecdotal feedback from staff since the launch. One of my favorites is from Nancy Lischer, Signal Theory’s Senior Talent Acquisition Manager: “I don’t know if you’ve done something to make it more humanlike in its interaction but I felt like I just had a full-on conversation!” We’ve also seen exponentially growing usage following the internal rollout in March.

AL’s chat interface has been a huge hit among our employees, now processing and generating close to half a million words a day. We found that employees were already seeking to use AI as a means to enhance their work. In fact, according to Microsoft’s 2024 work trend index, 75% of all employees are using AI at work. AL helps give Signal Theory employees an AI, tailored for advertising, that they can leverage without the data privacy and security concerns that come along with ChatGPT. This, coupled with internal training on how to best utilize AL, ensures we are well positioned to benefit from AI’s advancements moving into the future.

The Future

As more people with diverse experiences and expertise use AL, we will continue to explore new and exciting use cases. One that I am particularly interested in is the use of “impersonas,” essentially using automated synthetic survey data to extract scalable qualitative insights based on real quantitative data-supported personas. Early results indicate this data could be as much as a 95% match to real survey results.

Conceptually, I think it is fun to think about AL as an octopus. Intelligence of a different kind, weaving its way into every part of our workflow. AL plugs into data from existing first-party sources, allowing us to leverage it in new and powerful ways. It allows us to chase curiosities in an entirely different way and at a whole new scale.

If a system like AL’s is of interest to you and your organization, reach out to me at elijah.kleinsmith@signaltheory.com.