How Amplitude Ships Product Improvements in 1–2 Days with OpenAI

Wave is Amplitude’s proactive product agent, described as roughly 80% powered by OpenAI models including GPT-6.1 Sol, GPT-6 Luna, and GPT-Image-2.5 Flare. It combines Amplitude analytics, session replays, customer feedback, and code to identify opportunities, develop fixes, hand work to coding agents such as Codex, and measure the results. Amplitude estimates this can cut shipping time for product improvements from roughly a month to 1–2 days.

Teams give Wave a product area and the goals/metrics to improve. Different models handle different tasks: GPT-6.1 Sol drives the opportunity manager and deep investigations, GPT-6 Luna handles frequent low-cost work like scanning code and small evaluation checks, and GPT-Image-2.5 Flare produces mockups from session replay screenshots. Once an opportunity is ready, Wave sends a technical plan and acceptance criteria to a coding agent.

On Amplitude’s own homepage, Wave found that 80 of 100 sampled session replays had users typing their own questions while only 12 selected a generic suggestion card. It proposed personalizing the cards using recent charts, dashboards, or the originating page, while preserving free typing and fallback suggestions. After a human review, the change was handed to a coding agent and A/B tested; card-driven sends rose 160% relative to control after one day.

Amplitude’s model evaluations shaped these assignments. For the Opportunity Investigator, GPT-6.1 Sol scored 4.06/5 overall versus 3.62, 3.52, and 2.94 for alternatives, led all five investigation criteria, and had the fastest median investigation time at 54 minutes. In a separate Data Science Agent evaluation, it scored 67.5 versus 69.3 for another model but cost $1.28 versus $2.63 per dispatch, with higher scores on tool-contract compliance, estimation/uncertainty, and evidence grounding. For wireframes, GPT-Image-2.5 Flare scored 80.3 versus 63.9 across communication quality, product fidelity, instruction adherence, text/UI legibility, and visual coherence.

Customers across industries are already using Wave in beta, and teams can choose the level of automation: most send bug fixes straight to coding agents while bigger features get human review. Among customer-launched experiments, flags, and guides, one in four went live the same day Wave surfaced the opportunity, half in under four days, and two in three within a week; the source notes these milestones are interpolated and illustrative. Amplitude’s stated ambition with OpenAI is to make self-improving products the way companies build software.

How Amplitude ships product improvements in 1–2 days with OpenAI

View Original