Amplitude: OpenAI-powered Wave ships fixes in 1–2 days

Amplitude has built its business on helping teams see how people actually use their products, and Wave is its attempt to close the loop from insight to shipped change. The agent pulls together Amplitude data, session replays, customer feedback and code to spot opportunities, then drafts a technical plan and hands it to coding agents such as Codex. That division of labor is why Amplitude splits work across several OpenAI models: deep reasoning for investigations, cheaper models for high-volume checks, and an image model for mockups.

The homepage test shows how the loop works in practice. In a sample of 100 session replays, 80 users typed their own questions while only 12 selected a suggestion card, so Wave proposed personalizing the cards using recent charts, dashboards or the referring page. After review and an A/B test, messages sent through the cards rose 160% relative to control after one day. Amplitude's own evaluations also favored GPT‑6.1 Sol on investigation quality and speed, though a separate data science evaluation showed another leading model scoring slightly higher at a higher per-dispatch cost. Wave remains in beta, and Amplitude says most teams still route bug fixes straight to coding agents while new features get human review.