At Cannes Lions this year, Lily AI Co-Founder and CEO, Purva Gupta, joined a panel of visionary women to discuss the AI ecosystem in advertising and media today, including:
- Carolyn Pitt, Esq, Founder and CEO of Productions.com
- Verena Papik, CEO & Founder of Talkalytics
- Carolyn Murphy, GM Americas of Springboards
Below are 7 hot takes from the session at Salon Culture Conversations.
1. AI‑Enabled vs. AI‑Native
AI should deliver measurable business outcomes, not cool demo features that only work in unrealistically pristine data environments. AI-native companies understand the critical importance of training data for driving impact and building competitive moats.
Verena Papik, CEO & Founder, Talkalytics:
“What investors are really caring about… is to figure out how you can create a data set that’s valuable that will also give very specific insights to the industry, to the product or anything that you’re working on that’s giving you the competitive edge, but also at the same time enhances the customer experience.”
2. AI‑Native vs. AI Wrappers
AI wrappers get a bad “rap”, no pun intended. The truth is, every AI company has to differentiate its technology, even if they’re building on top of the same industry-leading models and LLMs. Training data is one differentiator; another is the use of feedback loops to continually refine and optimize model output for distinct, high-performing results.
Carolyn Murphy, GM Americas of Springboards:
“If we’re going to drive creativity forward, if we’re going to drive our industry forward, we need to think beyond the past. There are a lot of companies that are just wrappers on top of those (LLM) tools, and they’re just building off the past. One of the things that at Springboards we’re building is for variation because of that. That’s the difference when you’re talking about marketing strategy and brand strategy, you need that differentiator.”
3. Clean Training Data
Like many AI-native companies, Lily AI invested early on in a competitive moat: human-labeled, domain-specific data for model training.
Purva Gupta, Co-Founder and CEO of Lily AI:
“What Lily did when we started a decade back, and Silicon Valley gave me a lot of pain for it at the time, was focusing on really good training data for really good output. There was a lot of discriminative AI that we used where we taught the model with a lot of clean training data, as it required the right ingredients.”
4. Diversity by Design
Variation and differentiation of model output is influenced by the diversity of the humans training the AI. Creating diverse teams continues to drive higher-quality work, whether in AI or in business more broadly.
Purva Gupta, Co-Founder and CEO of Lily AI:
“More than half of my organization is female. More than half of my board is female, more than half of my leadership team is female. And that’s by design. It’s really important to make sure that we are intentional about some of these things because that diverse perspective will be really, really, really important in this new future because have access to a lot of tools.”
5. Emotional & Cultural AI Intelligence
AI must interpret emotional and cultural signals—not just raw data. “Culture as Catalyst” sessions underscored that AI tools need cultural fluency.
Verena Papik, CEO & Founder of Talkalytics:
“There’s a difference between, we talk about AI, there’s logical AI, and then there is the emotional AI. So the logical AI is really about the rational way we’re currently using GPT, but the emotional AI is truly also understanding the human input. What is the person feeling when it’s talking to me right now?”
6. Consumer‑Led AI
Unlike past technological advances, AI is not only broadly accessible but, in the case of Google’s AI Mode, almost unavoidable. Not only are consumers not complaining, consumers are embracing this technology for all of its conveniences and possibilities. This swift consumer-led adoption only accelerates investment in innovation, creating a rapid cycle of both disruption and transformation.
Purva Gupta, Co-Founder and CEO of Lily AI:
“This revolution is actually led by the consumer. This is not media type of a situation where the big tech is pushing it. Zero to a hundred million users per week for OpenAI happened in two months. So this is a whole different paradigm of how quickly the consumer is saying, I want this by noon Saturday. That’s how fast the consumer is adopting and moving.”
7. AI Upskilling
AI has infiltrated our personal and professional lives. Professionally, adoption isn’t an option; it’s a consumer and CEO mandate.
Carolyn Pitt, CEO of Productions.com:
“It’s about upskilling so that you understand what tools are being used today because next week it’ll be different tools. Years ago in law school, I learned about this Betamax case and years ago when the VCR was dropping, I don’t know if anybody remembers the VCR, there was a lot of fear out there. The major studios thought, oh no, this is going to take away financially from our returns. We can’t allow this to happen. So there was a Supreme Court case and ultimately if we fast forward, we all now know that when you have the opportunity to watch things on, nobody uses VHS but DVD and streaming and now it’s just multiple streams of revenue and opportunities that didn’t take away from the studios. It actually multiplied and gave more opportunity for them to generate revenue. That’s the way in which we have to start thinking about AI and other tools. The next AI, whatever that will be, you can’t stop it. You have to learn how to be fluid enough to work with it”
With that, those are the top 7 takeaways from a rich and fast-paced session that you can watch here!