May 31, 2023 · 27m · mad
An AI Assistant to Work Faster with Notion's Head of Data, Daniel Sternberg
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this Data Driven NYC episode, host Matt Turck interviews Daniel Sternberg, Head of Data at Notion, on how Notion built, scaled, and secured its AI features using flexible data infrastructure and user-centered design philosophy. Sternberg details Notion's multi-model architecture, enterprise privacy safeguards, and future vision for AI-driven workspace collaboration.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 12.2% of the talking time here. How this is scored →
Matt as informed peer 1.4 Guest teaching 3.4 Guest disagreement 0.1 Matt pushing back 0.0
speaking balance: gold is Matt, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 17:30 Polite factual correction on founding year
Daniel directly reframes Matt's earlier introductory statement by noting that Notion actually started in 2013 as a research project rather than in 2016.
Hardest push from Matt ▶ 13:24 Pressing for actionable enterprise preconditionsMatt pushes past general praise to demand concrete technical and organizational preconditions that companies must possess to deploy AI successfully.
Biggest teaching moment ▶ 17:30 Correcting company history timelineDaniel explicitly corrects the host's premise regarding Notion's timeline, explaining the deep context of the co-founders building in Kyoto starting in 2013.
Matt holds his own ▶ 9:58 Demonstrating knowledge of vendor architectureMatt cites specific industry reporting regarding Notion's dual integration with OpenAI and Anthropic to lead into a technical stack question.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Matt as informed peer | Guest teaching | Guest disagreement | Matt pushing back | Why |
|---|---|---|---|---|---|---|
| Notion's Data Infrastructure and Technical Stack | 2 | 3 | 0 | 0 | Matt asks a straightforward technical question about Notion's data stack. Daniel explains how they transitioned from vendor tools like Snowflake and Hex to building custom in-house pipelines on S3/AWS to handle scale. | |
| Overview of Notion AI Features and Use Cases | 1 | 3 | 0 | 0 | Matt congratulates Notion on its speed in launching AI and asks for an overview of capabilities. Daniel outlines core features such as automated page summaries, action item generation, and multinational document translation. | |
| Notion AI Technical Architecture and Development Strategy | 3 | 3 | 0 | 0 | Matt demonstrates contextual knowledge by citing reports of Notion using both OpenAI and Anthropic. Daniel details their multi-model strategy and explains how a cross-functional tiger team shipped the initial feature set quickly. | |
| Technical Preconditions and Scaling for AI Integration | 2 | 3 | 0 | 0 | Matt synthesizes the tiger team insight and asks what technical preconditions companies need to ship AI products fast. Daniel elaborates on how they repurposed existing data transformation pipelines and platform engineers. | |
| User Experience Design, Philosophy, and Internal Dogfooding | 1 | 6 | 1 | 0 | Matt asks about UI/UX challenges when deploying AI to general users. Daniel gently corrects Matt's opening remark regarding Notion's founding history, clarifying it was founded in 2013 rather than 2016, before explaining their intense internal dogfooding process. | |
| Future Horizons: AI Agents, Task Management, and Unstructured Data | 2 | 3 | 0 | 0 | Matt prompts Daniel for his view on hype versus reality in generative AI. Daniel highlights AI agents and bridging structured versus unstructured data in Notion databases as the most transformative opportunities. | |
| Audience Q&A: Data Privacy and Enterprise Security | 0 | 4 | 0 | 0 | An audience member asks about enterprise data privacy risks referencing historic Evernote policy backlash. Daniel thoroughly educates the audience on Notion's strict zero-logging policies, workspace opt-outs, and partner data processing agreements. | |
| Audience Q&A: Personal AI Wishlist and Human Collaboration | 0 | 2 | 0 | 0 | An audience member asks Daniel for a personal dream application of AI. Daniel shares his interest in personalized learning tools for his children and reducing tedious overhead in human workplace collaboration. |
Statements from this episode (12)
Sternberg: Few B2B SaaS companies match Notion's user behavior data depth
“We have an ability, I think, to look at The business and drivers of the business and tie them back to user behavior and customer behavior in a way that I actually think very few companies in kind of B to B SaaS do”
Sternberg: Notion's custom internal data infrastructure accelerated its AI feature launches
“As AI started to become more of a focus for us, some of these, some of this, ah, home built internal infrastructure that we were already starting to develop for some of those use cases, Started to be something that was actually helped us and kind of was a brid…”
Sternberg: Notion AI's top use cases focus on summarization and text improvement
“A lot of the most intensely used use cases of the current Notion AI experience are around kind of summarization, improving content.”
Sternberg: Notion uses models from both OpenAI and Anthropic
“While those are the two that I can say are, yes, publicly, like, we have leveraged both of those partners”
Sternberg: Notion began developing Notion AI before ChatGPT released
“While no one saw notion AI capabilities before ChatGPT, I'm not going to give us too much credit. It wasn't way, way, way before, but we were definitely working on this a little bit before that, and there was a pretty big focus at that point.”
Sternberg: A Notion co-founder actively codes and built early AI features
“One of our co-founders who still actively writes code was in his own world working on this for some time as well.”
Sternberg: Future AI software interfaces will not rely entirely on chat
“Over time, I actually think there will be a more radical transformation to the user experience. I don't think it's all, I personally don't think it's all chat all the time. It's like the way we're all going to interact with all products.”
Sternberg: Notion to launch first AI-enabled project management experience soon
“We are excited about some major improvements and kind of an all, a much more cohesive experience around how you manage projects in Notion. And some of that's going to be coming very soon. And it's going to be very much, I would argue, the first AI-enabled expe…”
Sternberg: LLMs' core transformative value is bridging structured and unstructured data
“One of the most interesting use, ah, most transformative use cases I think of generative AI and large language models is the ability to bridge between structured and unstructured information much more easily.”
Sternberg: Notion employees do not inspect external users' workspace content
“We do not actually, like, look at pieces of content from external users' workspaces.”