How Wave turns insights into deployable changes
Amplitude introduced Wave, a proactive product agent that combines product analytics, session replays, customer feedback and source code to identify opportunities and produce technical plans with acceptance criteria. Teams can send those plans to coding agents, including Codex, or review them manually before implementation.
The company estimates Wave can reduce the time to ship a product improvement from roughly a month to 1–2 days. In a test on Amplitude’s homepage, Wave recommended personalizing suggestion cards and produced a technical plan; after the change was A/B tested, card-driven messages increased 160% relative to control after one day, according to Amplitude. Jacob Newman, Principal AI Product Manager at Amplitude, described Wave as a unified platform of agents that “comb through massive amounts of session replays, analytics, and code to find opportunities to improve products.”
Model assignment and evaluation
Amplitude assigns different OpenAI models to tasks based on quality, latency and cost. GPT‑6.1 Sol is used for deep investigations and the Opportunity Investigator role; GPT‑6 Luna handles frequent, lower-cost checks; and GPT‑Image‑2.5 Flare generates wireframes and mockups from screenshots.
In Amplitude’s Opportunity Investigator evaluation, GPT‑6.1 Sol scored 4.06 out of 5 overall, leading on five investigation criteria and registering a median investigation time of 54 minutes. Amplitude reported a model cost of $1.34 for GPT‑6.1 Sol in that evaluation. In a separate Data Science Agent test, GPT‑6.1 Sol scored 67.5 out of 100 (compared with 69.3 for another leading model) at a mean model cost of $1.28 per dispatch versus $2.63 per dispatch for the other model. For wireframe generation, GPT‑Image‑2.5 Flare scored 80.3 out of 100 on measures including product fidelity and UI legibility.
Adoption and control
Amplitude said Wave gives teams control over automation levels: many teams route bug fixes directly to coding agents, while new features and larger changes receive human review. Among customer-launched experiments, flags and guides, one in four went live the same day Wave surfaced the opportunity, half launched in under four days, and two in three went live within a week.
Spenser Skates, CEO of Amplitude, said OpenAI models helped balance reasoning and cost across tasks and that improvements will let teams delegate more investigation and implementation work to agents as models advance. Amplitude published these results on November 12, 2026, and reported the engagement and evaluation figures above.
Original source: OpenAI News