Summary
Notes
Transcript
Offline. So last year on stage I teased Offline Mo to you guys and last month we shipped Offline Mo to all of you.
Well, since then, our subreddit traffic has been plummeting. Some Reddit users are wondering what to talk about, now Notion of Flammo is out. But to be clear, there's a lot more we need to do to make Notion and Notion Offline even better. And thank you again for your patience and for your feedback.
This is the reason we love our community so much, because you all care so much.
We also launched brand new AI products, like AI meeting notes. So, you can, thank you, my favorite too, yeah. Which does the note taking, so you can focus on being present in the meeting. And there's Enterprise Search. It turns your company's tribal knowledge into instant answers. We also launched Notion Mail.
on iOS and desktop. Android is coming soon. And last but not least, we upgrade Notion Calendar with scheduling link and task integrations. In today's keynote, you will learn how all these products, tools come together with Notion and with AI.
Last year was also a big year for our customers, with more companies and teams adopting Notion. Like Cursor. It's one of the fastest-growing companies of all time, and it runs on Notion. Or OpenAI. It also runs on Notion. They award thousands of new hire last year. And when those new hire have questions, they ask NotionAI. It's pretty neat to see that the company that start the current AI revolution rely on NotionAI.
And when you look at this new generation of AI companies, I'm really proud that almost all of them star on Notion.
So, outside the AI industry, Ramp is a fast-growing company reinventing finance, and I especially like their tech line. Time is money, save both. Last year, Notion helped REM consolidate half a dozen tools and save 70% of their tooling cost. And this is my favorite. To quote their CEO Eric, Notion is not just saving them money, but literally compressing time.
So now I think about it, or maybe Notion should steal Ramp's tagline.
Of course, we'll never forget small business and creators. Take Kai San Sai, a restaurant in Brooklyn. And according to the internet, they have the best pies in New York City. And I finally got to taste it this week. It's excellent. They're scaling and opening new locations.
And they're using Notion for almost everything. From training, to scheduling, to brainstorming new recipes. So, next time, when you have their pies, or thighs... Maybe there's a little bit of AI in it.
Notion is also growing in education, and this one is very personal for me. Some of you probably know Notion was reborn in Kyoto. Japan. We're inspired by Kyoto's arts and craft. And today, Kyoto University of Arts uses this notion across 800 faculty. It's very heartwarming to see that the school that teaches Kyoto's and arts and crafts now runs on Notion.
Thank you.
All the customer story comes back to our mission. Whether you're building an AI startup, reinventing modern finance, Or expanding your restaurant operations. We want Notion to be there. To support you with beautiful tools to help you build your life's work.
Notion, the company, is also growing. Last year, we opened five new locations. Sydney, Singapore, London, Paris, and Munich. So we can stay closer to our community and customers. In San Francisco, we also move into a new office. It's one of the first office buildings in the city, and it survived the 1906 San Francisco earthquake.
So if you zoom in to our building on the right, it says offices ready in May. So, right after the San Francisco earthquake, people were already eager getting back to office. So, return to office is different back then. So, time has clearly changed. So, you might be wondering, well, why am I talking about our office?
Because it relates to the theme of our keynote today. Around 1900, most Americans were farmers. But people began to migrate into cities for a new type of work, knowledge work. Office buildings were created to host knowledge workers. And new tools were designed to automate people's workflows. Tools like typewriters. So people don't have to second-guess their coworkers' handwritings.
Or file cabinets, which was basically the original second brain. And now imagine creating an entire YouTube course about file cabinets.
As we move into the software era, typewriters become more processors. File cabinets become databases and applications. And today, the software era became the software sprawl era. Tools are fragmented and rigid. Average company need to use ADA tools to run their business. And knowledge work feels more like busy work.
So sometimes we think, well, busy work is just modern life, you know, part of being an adult. Grow up!
But at Notion, our mission is to help you do your life's work. So we ask ourselves, can we use AI to help you delegate some of your busy work?
And as you probably know, AI and AI agent already exist for some sectors like customer support and coding. But the world has yet to see an AI agent for general purpose knowledge work. There are a lot of cool demos on Twitter, but nothing truly works at scale reliably. This is the problem we've been working on for the past two years.
And sometimes it feels everything we build at Notion has prepared us for this moment. So, what does it take to build a knowledge work agent? It needs three parts. First, it needs to know about your work. Now, context is everything for AI. But today, here's the dilemma of AI at work.
On one side, AI chatbot knows about your personal life, but it's not plugged into your work context. And on the work side, every SaaS app has AI now. But they don't work together, so the context is still fragmented. At Notion, we've been solving this fragmentation problem for years.
With Notion 1.0, we consolidate documents and knowledge base into one tool. And with 2.0, we're introducing databases and brought together tasks and projects into other workflows. Then, we have Notion Mail, Notion Calendar, and AI Meeting Notes. For everything else, there is Notion Enterprise Search.
This means Notion knows everything about you and your work, and we can pass this content to your Notion Knowledge Work Agent.
Second, to build a knowledgeable agent, it needs to be designed for teamwork. But most AI chatbot today are designed for single player. At Notion, we spent years building the essential foundation for teamwork with features like real-time collaboration, permissions, version control, and security guardrails.
Now, we can extend this foundation to the agent, and your Notion workspace becomes the place where humans and agents seamlessly collaborate.
Finally, a knowledge work agent needs to actually do real work for you. And this is the most tricky part. Because when your tools are rigid and your work context is fragmented, AI agents can't work reliably. They're just like people. They all struggle to piece everything together.
But Notion has all the work context, and Notion has all the core building blocks of knowledge work. So, we sent Notion AI to the bootcamp. And taught us how to do real work using those building blocks and contact in Notion. With a lot of hard work, and completely rebuilding Notion AI from the ground up multiple times.
We can say that it's finally ready.
And the result is, anything you can do with Notion, your Notion AI agent can do. It's the state-of-the-art AI agent that can do multi-step continuous autonomous work, sometimes 20 minutes at a time. This is such a big step change in our company's history.
And it gave us a brand new foundation to build our future.
So, we want to finally turn a page. Introducing Notion 3.0.
Thanks a lot.
It's the world's most advanced knowledge work agent, designed for teamwork. It's the first of its kind. We're really proud of it, and we can't wait for you to try it.
Before, Notion was your tool. With 3.0, Notion is your AI teammate. Before, Notion was your notepad. your knowledge base, your task tracker. With 3.0, Notion can do work for you. It can take your notes. It can give you the answer proactively. It can do your task end-to-end.
We think it will really reduce your busy work and give your time back for your life's work. Okay, let's see some demos. And here's sheer to tell you more.
Wow, what a great intro!
Thank you Ivan. I'm Shir and I work on AI at Notion. I've been working on AI for more than 10 years and it's never been more exciting than it is today. But, keeping up with AI comes with this. I don't like it either. So much of my reserved focus time is spent on things like
Writing status updates, prepping for meetings, updating docs. But recently, I've been able to reclaim more of my focus time. And actually, focus, because I've been delegating my busy work to this guy. This is our brand new Notion agent. It's being rebuilt from the ground up.
It's new and improved, and it can tackle much more complicated tasks, like this one you see right here. Previously, Notion AI could only do things one thing at a time. A single search, a single generation, you had to give very precise instructions for it to do the right thing. Today, the agent is able to take on a much more complicated goal or task.
Just like the ones you see here.
We have taught this agent to do everything that you can do in Notion and take advantage of the full set of building blocks. It can generate toggles, bullets, checkboxes, and my personal favorite, the table, columns. Anything you can do in Notion, this agent can do.
We're also introducing instructions and memory. This lets you control how your agent behaves. It's also a place for you to store notes on your preferences and memories about you. And, since it's just a Notion page, I can edit the instructions and memories however I want, and I connect it at mention to the rest of the context in my Notion workspace.
So, instead of telling you, let's show you. In today's demo, I'm going to be a PM working on follow-ups from our offline mode launch. First, I'm going to have my personal agent help with user feedback. And then I'm going to have it write a product requirements doc or a PRD that I can share with the rest of the team. Let's dive in.
As I get situated in my day, I'm going to first start with the task of gathering user feedback. I'm going to ask my agent to create a doc. Now, I know that there is a bunch of information from Slack, from email, and from the web, so I'm going to prompt the agent to look there.