The Hallway Track
21 tracked signals on personalization.
Spotify rebuilt its recommendation system as LLM-centric 'generative personalization' at 760M user scale
“moving from personalization as guesswork...to personalization as reasoning, which is able to analyze those results and really understand whether they are relevant to that user in a given context”
OpenAI launches ChatGPT Health feature integrating medical records and Apple Health data
Databricks launches CustomerLake, an agentic CDP on the lakehouse with LLM-powered profile and campaign agents.
“But now we can really personalize with really cheap, small, uh distilled LLMs, one-to-one uh continuous campaigns.”
OpenAI launches a new ChatGPT memory system that retains user preferences and context across conversations.
DoorDash distills LLM capabilities into traditional search systems using four semantic primitives
“the real bottleneck is semantic understanding. That is, it's knowing what elements actually mean to users, given a certain context, and what the shopper actually intends to do”
Adobe is building real-time AI-personalized websites using LLMs and RAG grounded in site content
“We don't want the whole site to be generated. I mean if you talk to marketing people they have a very strict brand guidelines. You don't want to just come up with our have some hallucinations there.”
Early ChatGPT memory forced users to manage memory creation, burdening rather than helping them.
“as a user, because you could see every time a memory was created, it felt like you were responsible for both creating memories while you were just trying to have a conversation”
Circles telco saw 22% ARPU increase and 9% churn reduction using OpenAI API and Codex
Databricks launches CustomerLake, an agentic customer data platform embedded natively in the lakehouse.
“It can power one-to-one personalization with agents to provide the perfect customer experience a billion times a day.”
Databricks proposes 'Infinity Campaigns' where a per-customer agent continuously generates personalized next-best-action content in a feedback loop.
“What if we could have an agent for every customer?”
Windows now lets AI agents control OS personalization (wallpaper, accent color, RGB lighting) via natural-language skills defined in a skill.md file.
“That markdown file is your contract with the agent. It declares what tools are available, what commands to run, and what parameters they accept.”
Amazon Payments achieved single-digit conversion lift using contextual multi-armed bandits on SageMaker AI
“The problem turned out to be the content, not the model.”
NVIDIA Dynamo-Triton enables deployment of HSTU generative recommender systems at scale
Virgin Atlantic uses Databricks and Unity Catalog to deliver real-time passenger personalization to inflight crew.
“we're using the Databricks platform to get that information to our crew where they can apply that extra context, that Virgin magic to make sure that you don't get the gin and tonic, you get the champagne that you really want.”
Instacart envisions a unified in-store/online shopping mode powered by a continuously learning AI system within 5-10 years.
“in 5 to 10 years, customers shouldn't have to think about shopping in store or online. It will be one single unified mode powered by this continuously learning AI system”
Stanford researchers are building human-centered AI tools using music emotion recognition to aid people with depression and anxiety.
“we are creating AI tools that will use music to aid people with depression and anxiety.”
Databricks combines Lakebase and AI Search to power real-time e-commerce personalization.
Discovery Bank uses behavioral AI and real-time decisioning for hyper-personalized banking at scale
Preply uses OpenAI to power AI-generated lesson summaries and personalized language learning feedback.
Personalized outreach is essential for successful outbound sales.
“Spend time thinking about one individual person. What is her problem specifically?”
F1 uses Salesforce to personalize experiences for 831 million fans at scale
“Salesforce helps us bringing all those data points together so we really understand our fan across all their experiences within the ecosystem.”