Amplitude says Wave agent can cut product-improvement time from about a month to 1-2 days

Amplitude says its proactive product agent, Wave, which it describes as roughly 80% powered by OpenAI models, can help teams ship product improvements in one to two days instead of roughly a month.

Wave combines analytics, session replays (recorded user sessions that show how people use a product), customer feedback and code to identify opportunities, produce a technical plan and then hand that work to coding agents. On Amplitude's homepage, a Wave-led personalisation of suggestion cards increased messages sent through those cards by 160% relative to control after one day. In a sample of 100 session replays, Amplitude reported 80 users typed their own questions and 12 selected a suggestion.

Amplitude tells OpenAI's customer story that it assigns different OpenAI models to different tasks: GPT-6.1 Sol for deep investigations and opportunity management; GPT-6 Luna for frequent, lower-cost scanning and checks; and GPT-Image-2.5 Flare to generate mockups and wireframes. In its internal evaluations, Amplitude reports that GPT-6.1 Sol scored 4.06 out of 5 on Opportunity Investigator criteria and had a median investigation time of 54 minutes. In a separate Data Science Agent evaluation Amplitude reports GPT-6.1 Sol scored 67.5 out of 100 at a mean model cost of $1.28 per dispatch versus $2.63 for another model. Among customer-launched experiments, Amplitude says one in four went live the same day Wave surfaced the opportunity, half went live in under four days, and two in three within a week.

"The idea with Wave is you have a unified platform with agents that comb through massive amounts of session replays, analytics, and code to find opportunities to improve products," said Jacob Newman, Principal AI Product Manager at Amplitude, in Amplitude's case study on OpenAI's site. "It is a great example of how non-groundbreaking things that just get ignored can increase the adoption of features in unexpected ways," said Eric Carlson, Senior Principal AI Engineer at Amplitude, in the same case study.

For African product teams, the headline speed and cost gains Amplitude reports depend on three operational facts beyond model quality: where the models and inference run, where customer data can be stored, and which cloud accelerators ( specialised inference hardware such as GPUs and TPUs) are available nearby.

OpenAI says regional data residency is currently available in Europe, the United Kingdom, the United States, Canada, Japan, South Korea, Singapore, India, Australia and the United Arab Emirates. That list does not include any African country today, so customers in Africa who need in-region storage would have to wait for OpenAI to expand residency or store data outside the continent under whatever cross-border rules apply locally.

Cloud providers also limit where specialised inference hardware and model hosting are available. Microsoft's Foundry documentation notes that model availability varies by region and cloud, and Google Cloud's accelerator listings show that the availability of specific GPU and TPU types varies by region and zone. Some high-end accelerator types are available only in a small set of locations. Those regional limits affect latency, throughput and cost for customers who must send requests to distant regions to use certain models or accelerators.

That combination matters for African firms weighing agent-driven workflows: if Wave or the specific OpenAI models it uses are only offered from outside Africa, teams may face higher latency, cross-border data compliance steps and added cloud egress and accelerator costs. Important questions for African customers include whether Wave and the models named by Amplitude (GPT-6.1 Sol, GPT-6 Luna and GPT-Image-2.5 Flare) will be available to run and to store data in-region, and whether the cloud accelerators used for inference will be deployable in African regions.

The Amplitude/OpenAI case shows how much agent-driven development can speed delivery, and why the practical value of that speed for African businesses depends on where models, data residency and accelerators are located.

Newzroom dot Africa has reached out to Amplitude and OpenAI for comment and is awaiting a response. This story will be updated when they respond.

Source: OpenAI