Amplitude

Every AI-generated UI has the same tell. Generic color palette, awkward spacing, the unmistakable look of “an LLM made this.” William Newton fixed that by baking Amplitude’s design system directly into a Claude agent, built on Cloudflare in two days. 2,219 sessions in the first few weeks. Here’s how he built it. Read more: https://bit.ly/4wKKlbE

Should your agent have a personality? Ours does, and we designed it on purpose. Here's how you can too ⤵️ 1. Pick the traits that matter for your product. For us, that was inquisitiveness and helpfulness. 2. Ship a first version on intuition. Treat your design choices as hypotheses, listen to what users say, and watch how the agent performs. 3. If something feels off, size the issue before changing it. It’s only a real signal if you know how often the pattern shows up. This is the step most teams skip. 4. A/B test the fix to see whether your change solved the issue. Jacob Newman shares the full learnings: https://bit.ly/3R6grOE

Churn is the symptom. Better experiences is the cure. Glenn Vanderlinden of Human37 shares his strategies for fixing user experience problems and how he implements them with Amplitude. Read more: https://bit.ly/4dANTnY

If someone's been clicking around your website and hasn't converted yet, should you say something? Real-time targeting in Web Experiments lets you find out if showing a relevant experience the instant someone qualifies can improve your conversion rates. Nithya Rajendran, Product Manager at Amplitude, shows you how to get beta access to real-time targeting.

Customer feedback just got a lot harder to miss. Amplitude Guides and Surveys now supports Slack alerts for survey responses — route feedback directly into the channels where work already happens. Product ideas to Product. Website feedback to Growth. Critical issues to Support. Available now for Growth and Enterprise customers, including EU. https://bit.ly/43hlmPo

Run your next feature launch from one Slack channel with Amplitude AI Agents: → Ask Global Agent how the new flag is trending vs. control. → Ramp or roll back the flag without looping in eng. → Pull in session replays when the metrics drop. → Get the post-launch readout with both qual and quant insights. Read the full workflow: https://bit.ly/4uxNhXD

Most product fixes are patches. You find the broken thing, fix it, and six months later something adjacent breaks. Heatmaps needed a full rebuild. Screenshot capture, element placement, device handling, URL targeting — the problems were too interconnected to address one at a time. So we did. The result is a heatmap that matches your live page and data you can act on without first wondering if the tool is working correctly. Read more: https://bit.ly/4dcSYnt

Yikes sorry for the delay, Claude just finished this infographic. Here's what our teammates voted for as their longest task time during AI Week.

Shipping faster isn't the hard part anymore. Validating whether what you shipped was actually good is where teams actually get stuck. That's the problem experimentation addresses head-on. As AI coding tools accelerate output, the gap between shipping and validating has grown wider, not smaller. Eric Metelka shares how Amplitude and Statsig are building the validation stack that closes that gap with offline evals, progressive rollouts, online experimentation, and LLM observability working together as one system 💥 Read it here: https://bit.ly/4uIYWlL