When businesses talk about becoming “AI-first,” they often just mean adding AI features to their products. For us at Amplitude, that’s certainly part of it (check out our )—but beyond what we build, becoming AI-first is about changing how we build. AI has to be engrained in our daily practice.
That’s why we decided to run an AI Week. Over a five-day sprint, our product development org hit pause on business-as-usual and came together to accelerate AI learning, shift mindsets, and apply AI in meaningful ways to real challenges. This wasn’t just another hackathon—it was an investment in building our AI muscle.
Here’s what we did, why it worked, and what surprised us most. And if you want to try it, what you and your team can do to make your own AI Week a success too.
Why AI Week?
AI is already reshaping our industry. But most teams are stuck in a passive mindset—watching, dabbling, waiting. I felt like if we really wanted AI to impact how we build at Amplitude, not just what we build, we needed to create the space and the pressure to learn by doing. We needed to dive in.
AI Week let us get hands-on with a new technology, see what we could build with it, and reflect on how to excel. By bringing together engineering, product, and design, we got all development teams to commit to collaboration and align on a shared goal: move fast, learn faster.
What made it work
- Doing the prework. In the weeks leading up, we shared our AI strategy, expectations for the event, and a curated AI fluency starter pack (tools, articles, and internal agents) across the org. Everyone had a baseline understanding of what we were doing and why.
- Building in person. We worked side by side at whiteboards, in the hallways, and huddled around screens. Face-to-face collaboration kept spirits high and prioritized the human side of learning.
- Anchoring in real deliverables. Teams pulled forward real projects—work that was already on the roadmap—so they weren’t building toy examples. Applying AI to concrete problems made the work meaningful.
- Modeling the how. Curiosity, speed, vulnerability: we lived the behaviors we wanted to see in our peers. As we explored unfamiliar tools, we asked naïve questions, took risks, sought guidance from others, and most importantly, didn’t wait for permission.
- Leading by example. This wasn’t something leadership greenlit and disappeared from. We showed up, built, learned, and shared successes and failures alongside their teams. That set the tone: this was a priority.
What the week looked like
We designed the week for flow, balancing hands-on work with just enough structure to keep everyone stretching, sharing, and finding pathways to success. Each day built on the one before it to move the org from awareness, to fluency, to application.
- Day 1: Grounding. AI strategy, an external expert panel, hands-on AI demos from senior leaders, and a deep dive into promising use cases.
- Day 2: Training. Tool-specific sessions on , , and our in-house AI tool Moda, followed by discipline-specific workshops for engineering, PMs, designers, and support teams. Teams then “called their shot” by pitching the projects they’d tackle the rest of the week.
- Day 3: Building. Cross-functional triads (engineer, product, designer) dug into their work. Mid-day, we met for Show, Tell, Ask sessions to demo progress, share lessons learned, and ask each other for help.
- Day 4: More building. Same as Day 3—more growth, more sharing, more momentum.
- Day 5: Building + Celebration. A final day of deep work, capped off with awards for the Best AI Breakthrough, Best Prototype, Best Call-Your-Shot, and of course, Best AI Fail.
And yes—we ended with a happy hour.
What surprised us
Two kinds of outcomes stood out, one qualitative and one quantitative.
First, the personal impact. People told me this week was a wake-up call. That they felt like they weren’t just learning a new skill—they were investing in their own careers and realizing they could grow with AI instead of fearing it.
And second, the productivity boost. We measured over a 40% jump in productivity for teams that embraced AI in just the first week. Several teams told me they pulled off in five days what they’d originally scheduled for multiple weeks the following quarter. One team estimated they cut a three-month project down to three weeks.
And then there were the ideas. Teams demoed an interface that lets you talk to customer data and generate the exact chart you need. Another built a way to generate an entire sitemap just by describing the outcome. That kind of innovation is what happens when you give talented people new tools and the time to explore them together.
What’s next?
AI Week wasn’t a break from work—it was the work. It made the abstract tangible, the skeptics curious, and the builders unstoppable.
Now the real transformation begins. Going forward, our whole org has the shared experience and understanding to put our AI skills into everyday practice for continued gains in productivity and innovation.
We’re also using the second half of our 2025 roadmap as a proving ground. We’re excited to debut more AI features in Amplitude soon, including some from AI Week, and pass the benefits of AI on to our users too.
My advice for other leaders
Your strategy is only as strong as the investment you make in it. If you’re serious about becoming AI-first, I’d encourage you to make this one. We hit goals faster than we thought possible and created a whole new level of energy around how we build.
A few final takeaways:
- Commit personally. Leaders have to do more than say, “Go do AI.” Show up and do the work too.
- Prepare ahead. Give teams some baseline prep so they come in ready to go.
- Hold a high bar. Don’t water this down into generic exercises. Push people to work on real problems that matter.
- Make it okay to fail. Highlighting our AI fails made it easier for people to take risks.
- Prioritize in-person. Getting together face-to-face created trust and made it fun.
If you’re doing something similar and want to trade notes, reach out. I’m happy to share more.