By Rick Byrne
COO and Co-founder, Pluralytics
As a first-time founder, I have developed a certain appreciation for Elon Musk’s often quoted and paraphrased admonition of what it is like to manage a startup: “Running a start-up is like chewing glass and staring into the abyss. After a while, you stop staring, but the glass chewing never ends.”
Three years after launching Pluralytics, I think “glass-chewing” is apt because it is an analogy for a fundamental driver of startup life: It is the hard, never-ending problem solving that is the founder’s motivation. I think Elon would agree. What else would account for the phenomenon of the serial entrepreneur?
I think most founders would say problem solving is not just a condition of startup life, it is a key part of the job where you find joy. If it isn’t, then you might consider that before beginning the more onerous work of raising money based on your idea and your articulated prediction of the vast and unknowable future, otherwise known as the abyss.
Since our launch, the founding team at Pluralytics has found a particular joy in solving problems. Our “prediction” has been that using AI to understand values signals in language can help marketers and communicators deepen engagement with their target audiences by identifying the language that is most likely to connect. We’ve been doing this with our own proprietary language model, which has been operating and being used by our Fortune 500 customers years before OpenAI made its big splash with ChatGPT last fall. We are as fascinated as – or perhaps more than – the rest of the world with the power of NLG, and we have been working with OpenAI since May 2021 as part of their early beta to uncover, parse, and generate language signals.
Recently, as part of our ongoing efficacy testing, we saw once again the power of values to predict and improve language appeal. Using Momentive’s survey platform, we clustered consumer targets by values attributes (our 5 proprietary values personas) and delivered a split test with a control and a values-driven rewrite to measure response. We used our unique methodology, which produces an index score to calculate potential language appeal and tested it with values panels.
Using publicly available marketing messages from New York Life, Charles Schwab, and Unilever, we employed our system, which will work with any LLM, to automatically rewrite versions so they scored high for appeal to each values segment. We have been using language analysts to rewrite copy using an internal suggestion algorithm for our customers since we launched, but this was our first test with copy that was auto-generated and leveled-up to over-index for each target segment – our differentiator.
What did we find? Pluralytics AI appeal-boosted language was chosen 33% more often by targeted consumers. This result is consistent with our previous testing with our human-in-the loop service and with what our customers tell us is a key benefit of our ValuesFinder SaaS platform: deeper engagement. Target audiences preferred our values optimizations 2.5 to 1 over the controls.
High performing instant rewrites create blue sky possibilities for our future, which we are currently exploring and building. Up next? Instant values language translation at scale delivered through APIs and plug-ins to power and improve language anywhere it exists, using any LLM.
For tech platforms that help organizations manage content – think digital asset management, document management software, social media management – we believe we have a solution that provides high value for companies seeking to accelerate their AI/NLG product roadmap and avoid falling behind.
For founders like us, the incessant act of problem-solving is our joy and is occasionally punctuated by the wins along the way. If you are lucky, this happens more often than not. Either way, it is time to chew some more glass.
Want to accelerate your AI roadmap with language analysis and rewrites that perform? Contact Rick Byrne at email@example.com to be considered for Pluralytics’ closed beta for APIs and developer tools.