Hello, Joyce! Take us through your career experiences and early AI encounters that excited you about the technology, gradually leading you to the position of Head of Generative AI at Amperity.
I’ve always been fascinated by how people can make more informed decisions, which is what drew me to statistics and ML. I started my career in consulting, and then joined Custora, a customer analytics platform, as a founding member of the product team. Over a five-year period, we were able to grow Custora from an MVP to a product used by seven of the top 20 brands by ecommerce revenue before we were acquired by Amperity in 2019.
Since then, my focus has been AI, ML, and ML Ops, in the service of helping brands unify, manage and activate their online and offline customer data to build a first-party strategy.
Over the last year, we’ve realized there was so much value we could provide to customers and I stepped in to lead our Gen AI efforts. Gen AI will democratize data access, unlocking more data-driven decision making throughout the enterprise. Our Customer Data Platform (CDP) with AI will make personalized experiences a reality, and give brands the opportunity to lessen resource constraints in the realms of creative and customer service.
What are your responsibilities as the Head of Generative AI of Amperity? Also, how does your role align with and help accomplish the overarching company goals?
Amperity is on a mission to equip brands with the ultimate tool: the world’s finest unified customer profile. This profile is designed to supercharge analytics and energize a variety of applications.
In my role, I’m at the helm of steering our product development and strategy for all our gen AI initiatives. It’s a role I love, as it allows me to juggle a diverse array of responsibilities, from the nitty-gritty technical details to crafting our go-to-market strategies.
Some of the ways that we’re using AI to support our mission include the following:
-Data democratization. Imagine if everyone from the CEO to an analyst could make smart, data-driven customer-focused decisions. That’s what Gen AI has the potential to enable. It lessens the barriers to data access. No more wrestling with SQL, getting lost in complex data structures, or scratching heads over visualization tools. It’s all about making data friendly and accessible to everyone.
-Ecosystem enablement. Here Gen AI acts as a creative superpower. It drives the marginal cost of content creation to zero, enabling a host of use cases across personalization, creative generation, and customer service. But for these use cases to be successful, having a robust unified customer data foundation is a prerequisite.
-Engineering velocity. With Gen AI, engineers are able to innovate faster than ever, making problems that were previously impossible to solve tractable. We’re using it to deliver value to our customers more quickly.
How do you approach the challenges and opportunities of building and deploying generative AI models for marketing applications?
When developing Gen AI products, I adhere to three guiding principles:
– Selective Use of Gen AI: Gen AI is a powerful tool, but it’s not always the first one I reach for. I start by deeply understanding the customer problem we’re addressing. This involves asking if Gen AI is truly the best fit for the solution, and exploring alternative non-AI methods for solving the problem, along with their pros and cons compared to an AI-based approach.
-Data Availability and Accuracy: The success of any AI or ML project hinges on the quality of the data used. Especially in marketing applications, the key lies in having a precise and up to date customer profile. Ensuring we have the right data is always my initial step, setting the stage for effective AI application.
-Rapid Deployment: The field of Gen AI is in a state of rapid evolution, with new models and capabilities emerging continuously. Given this dynamic environment, the best practice is to launch products swiftly to receive immediate user feedback. This approach involves being open about the limitations and developmental aspects of the product. Additionally, it’s important to differentiate between the challenges our company is uniquely equipped to tackle and those likely to be addressed by larger players in the industry. This distinction helps in focusing our efforts where we can make the most impact.
Amperity uses patented AI/ML methods to stitch together rich customer data across disparate sources. How do you approach the development of these AI/ML methods, keeping in mind the market requirements and especially the context of customer data?
ID resolution is solving the problem of really knowing your customer. Customers interact with brands across so many touch points from online to brick and mortar to email to loyalty programs. When developing our ID resolution there were a couple principles, we kept mind:
-Comprehensiveness: Amperity is able to take in and stitch diverse data sets from transactions to email engagement to web events. We’re able to provide a unified and accurate customer profile across all of a brand’s touchpoints.
-Transparency: We output a confidence for each stitched record and our users are able to see why particular records were stitched. This builds trust with end users.
-Freshness: Identity resolution is only helpful if it’s up to date. Our stitch algorithm runs on a daily basis so brands always have access to the most up to date customer profile.
Generative AI is still a technology in the making, and its entire capabilities are still to be explored. How do you plan to further blend this growing technology into Amperity’s products, making them more advanced while staying within ethical boundaries?
While Generative AI is a newer technology, AI is not new for Amperity. Since our founding in 2016 our patented identity resolution process has been rooted in AI / ML. And we continue to invest in AI capabilities that we know will help drive our customer’s business goals forward and empower them to truly understand their customers.
We’re doing this by developing new capabilities that use the power of Gen AI to support the users of our platform. This uses AI to develop and iterate on queries and segments, and to enhance the ability to expedite customer data exploration within the platform. Our investment decisions are significantly guided by customer feedback and we’re planning to keep the conversation going with them. Their feedback is like a compass for us – it tells us where we should be heading with our next round of improvements.
Finally, as an increasing number of brands integrate generative AI into their strategies, the quality of their underlying data assets will emerge as a key factor setting them apart from competitors. This underscores the importance of having a CDP to efficiently capture, manage, and use this data. Additionally, ensuring the data is first-party and obtained with consent mitigates the risk of using non-consented data in AI models, thereby aligning with privacy standards.
How do you see AI transforming the very fabric of marketing in years to come?
The fusion of readily available data and AI is poised to initiate a golden era in personalization, enabling the delivery of exceptional, tailor-made experiences on a massive scale. We are now approaching a point where AI can mimic the personalized, one-on-one interactions offered by a store associate, while simultaneously offering the convenience and breadth expected from digital shopping.
In the realm of marketing, this evolution is set to revolutionize content creation. AI will be capable of autonomously producing creative content for advertisements, using customer data like demographics, interests, and shopping habits. This will not only allow for hyper-personalization at scale but also enable leading brands to shift their focus to more impactful work.
AI will also lead companies to a more data-driven approach. Data will become more accessible and simpler to interpret and apply in decision-making. The need for specialized skills like SQL expertise or an in-depth understanding of data schemas will no longer be a barrier to data access.
What are the DOs and DONTs that organizations and marketers should be conscious of?
Do:
-Be explicit about the specific use cases where you plan to use data and AI and specify the expected outcomes. What results do you expect to achieve?
-Carefully evaluate if Gen AI is the most appropriate tool for your specific use case.
-Prioritize data quality and comprehensiveness – establishing a unified customer data foundation is essential for an effective AI strategy.
Don’t:
-Rush to implement Gen AI across all areas. Start with a manageable, human-in-the-loop use case, such as generating subject lines.
- About Joyce Gordon
- About Amperity
Joyce is the Head of Generative AI at Amperity, leading product development and strategy. Previously, Joyce led product development for many of Amperity’s ML and ML Ops investments, including launching Amperity’s predictive models and infrastructure. Joyce joined the company in 2019 following Amperity’s acquisition of Custora where she was a founding member of the product team. She earned a B.A. in Biological Mathematics from the University of Pennsylvania and is an inventor on several pending ML patents.
Amperity is the customer data platform that helps brands build the unified first-party customer data foundation they need to grow revenue, reduce costs, save time, and stay compliant.
Our unparalleled solution creates accurate, reliable profiles by stitching together every type of rich customer data across disparate sources using patented AI and Machine Learning methods.
To know more about: www.amperity.com