Amplitude leverages OpenAI models to accelerate product improvement cycles to days

Amplitude has introduced Wave, a proactive product agent built substantially on OpenAI models, which streamlines the process of turning product insights into shipped improvements. The system integrates Amplitude data, session replays, and customer feedback to identify and prioritize opportunities, coordinating implementation with coding agents like Codex. Amplitude estimates this process can reduce the typical development time from about a month down to just one or two days. Wave utilizes specialized OpenAI models for different functions; GPT-6.1 Sol handles deep investigations, while GPT-6 Luna manages lower-cost, high-volume tasks. In testing on its own homepage, Wave suggested personalizing suggestion cards, leading to a 160% lift in card-driven sends after one day, as noted by Eric Carlson, Senior Principal AI Engineer at Amplitude. Evaluations show GPT-6.1 Sol achieved a 4.06 out of 5 score in the Opportunity Investigator evaluation, and in the Data Science Agent evaluation, it achieved a mean model cost of $1.28 per dispatch. Jacob Newman, Principal AI Product Manager at Amplitude, stated that the platform allows agents to comb through massive datasets to find improvement opportunities. For high-volume tasks, Amplitude uses GPT-6 Luna to maintain low latency and cost.