Amplitude AI Agents Demo

Quick Overview

Amplitude's AI Agents streamline complex data analysis tasks, such as investigating a loyalty program's low conversion rate (5.42%) and quickly identifying that existing customers convert at nearly twice the rate of new users due to in-box discount coupons, which the AI agents can analyze, visualize, and suggest next steps for, all within minutes.

Key Points: The AI Agent successfully investigated a loyalty program conversion funnel, finding a low 5.42% conversion rate from the welcome page to enrollment over the last 30 days. The initial analysis revealed that existing customers convert at nearly 2.44 times the rate of all users (14.2% vs 5.51% conversion), indicating a dramatic difference in engagement. The AI attributed the higher existing customer conversion to the 'in-box discount discovery,' where 30% of loyalty program joins came from customers finding discount coupons inside their delivered grocery boxes. The agent automatically pulled supporting session replays linking specific user behavior (entering the 15% off coupon code) to successful enrollment. The AI agent synthesized quantitative data (charts, conversion rates) with qualitative data (customer feedback snippets) to explain the 'why' behind the conversion gap. The AI agents can handle various tasks including dashboard analysis, session replay analysis, customer feedback analysis, and website conversion optimization experiments. The process, from asking the initial question to receiving synthesized answers and suggested next steps, takes mere minutes, replacing time-consuming manual analysis.

Context: The video features Janaki, an AI Systems Engineer, demonstrating the capabilities of Amplitude's AI Agents, which are designed to accelerate data analysis, experimentation, and insight generation. The core demonstration focuses on diagnosing a low conversion rate within an online grocery store's loyalty program, contrasting the performance of new users versus existing customers to uncover the root cause of the drop-off.

Raw markdown version of this recap