Through LiveRamp’s corporate venture capital fund, LiveRamp Ventures, we get to work with some of the most cutting-edge companies at the forefront of customer data infrastructure and applications. We’re excited to collaborate with the next generation of these companies, and believe that they will drive tremendous value not just for our clients, but for the entire industry.
One such company is Frame AI, which helps to apply machine learning and natural language processing to unstructured customer feedback. LiveRamp sees applying AI, machine learning, and natural language processing to customer feedback and using this to enhance customer data as a critical step towards addressing one of the fundamental problems we’re collectively working to solve, with some of today’s cutting-edge tools. Many have talked about the ability of AI to generate content, but what savvy stakeholders are realizing is that AI can be just as helpful with analyzing and actioning on the mind-boggling amounts of customer data that already exists, as well.
We asked Frame AI’s co-founder and CEO, George Davis, about some of the most burning questions we had about AI right now. George is no stranger to AI, having applied it to genomics, disrupting human trafficking, and predicting M&A activity based on SEC filings as part of his PhD. Immediately prior to Frame AI, George was Head of Adaptive Learning at Knewton, where he led a data and engineering team driving educational experiences for 10 million students around the world.
Q: You named your company “Frame AI” years ago, so clearly you’ve been a believer for a long time. What was the AI opportunity you saw then, and how has it changed?
Frame actually got inspired by a trend that has little to do with AI. Building customer relationships in the smartphone era means finding many ways to communicate – voice, SMS, webchat, social, communities, etc. – because businesses must compete with everyone who will go that extra mile. And yet, even brands that are great at reaching customers can benefit from help with listening to them. Talking without listening leads to lost trust and runaway costs that get worse with scale – and makes consumers’ lives worse.
As ML and Data consultants at the time, we saw the same “why” driving this problem at dozens of companies: customer interactions were being recorded by different teams with different tools and different goals. When engineers map that data into a CDP or Data Lake to build “golden profiles”, they often ignore the most valuable stuff – organic communication customers took the time to write or say – because it is hard to work with.
So, businesses and customers both invest so much in making contact and holding conversations, but the data just sits there, crackling with potential energy – both in terms of learning about customers and figuring out which customer ops are actually good investments. And we saw a “how” that could fix it: AI, especially Natural Language Understanding, was in the early stages of a revolution that would let us actually extract the important facts from those conversations.
So that’s what Frame AI does. We help businesses extract facts about customers (and their own operations) from call transcripts, social posts, reviews, support tickets, and other interactions. Then we weave those facts into clear metrics about what’s driving your costs and revenue outcomes, richer customer segmentation, and automation that powers more proactive and efficient customer ops.
The main way that mission has evolved over the years is that more stakeholders have gotten involved. Early on, we worked primarily with support leaders trying to understand and reduce their costs. But over time, product teams, marketing teams, and the full revenue lifecycle have woken up to the idea that they need to maximize value from their most unique asset – the time their customers will spend with them.
Q: We’ve been through cycles of hype around AI before. Is this time different?
Hype cycles are driven by unbounded potential on one hand, and fear of missing out on the other. The combination can make people overinvest in specific ideas about how a new technology will be useful. In that sense, the current cycle has the same dangers.
The giant difference is that this hype is not about some unproven, “almost here” future. Large language models (LLMs) – machines that can understand, compose and act on human language – are here. Thanks to OpenAI, the cutting edge is immediately accessible to anyone with an internet connection, and is manifestly useful for purposes ranging from entry-level research to enterprise-scale data analysis. Adoption has outpaced the hype.
We’ve been using large language models at Frame AI for years – essentially since the term was coined. But even as believers in this tech, we were caught unaware by the sudden accessibility of the latest generation, and how rapidly they are being adopted as a result. This is going to change how many jobs work, and what consumers expect in terms of interaction and accessibility.
We talk about it as a 30,000 foot wave, because it should prompt folks to reconsider fundamentals about how they describe their market, products and roles.
Q: It sounds like you’re suggesting people be concerned about everything, all at once! Is this a giant risk for businesses? For marketers’ jobs in particular?
AI is a dangerous trend to ignore, because your competition will not. However, the opportunities FAR outpace the dangers. Today’s businesses slog through analyzing data and automating routine tasks, often ignoring opportunities or taking on expensive, mind numbing tasks because resources aren’t available to adapt those tasks to machines.
Modern AI is eliminating the barrier to interfacing with machines. Through companies like Frame AI, it’s making it possible to extract useful signals from messy data (like human speech) without weeks of engineering. And through companies like OpenAI, it’s making it possible to automatically reorganize and present the key facts from that data as appropriate for specific tasks. All without new code.
Anyone can make the most of this moment by going back to fundamentals. What are your goals? And what are the resources – data, distribution, etc. – that make you uniquely capable? AI can’t give you purpose or invent value out of thin air, but it can help you bring what already makes you unique – as an individual or a business – to market much, much faster.
As an individual, what makes you unique might be your ability to draw out important insights from a peer or a customer in conversation. But AI can help you turn the transcript of that conversation into a summary that makes that valuable accessible going forwards. As a company, one thing that makes you unique is your surface of interactions with your customers – every sales call, every support ticket, every review. Frame AI can help turn that data into actionable priorities for your operational and product teams, or new segmentations for your marketing.
For more about working with LiveRamp, emerging risks of AI adoption, data access, AI and the enterprise data stack, and more, check back for Part 2.