Wave

Wave is Amplitude's team of proactive AI agents for discovering and prioritizing product improvements.

It takes on the work of finding and connecting evidence across your product data, session replays, and customer feedback, helping your team uncover improvements that might otherwise go unnoticed. It brings its findings together as prioritized opportunities, each with supporting evidence, a proposed solution, and target metrics. Your team spends less time gathering context and more time deciding which changes to pursue.

Availability

Wave is a paid offering for organizations on Growth and Enterprise plans. To get started, bring a business goal and talk to a Wave expert. The Amplitude team helps you assess your product analytics and Session Replay data, and shows how Wave can help you investigate opportunities, ship changes, and keep learning from the results.

How Wave works

Set the focus for Wave around the metrics you want to improve, review the opportunities it finds, and choose which changes to pursue.

Define the focus. Tell Wave what you want to improve, such as your checkout experience or activation rate. Its agents help create a product area with relevant goals, target metrics, and boundaries. Then you can tune and configure product area settings to guide Wave toward recommendations that match your priorities.
Discover opportunities. Wave maps your product and sends agents to investigate hotspots. You can also direct it to investigate a specific question. With scheduled investigations, Wave uncovers new areas to explore and goes deeper where the signal calls for further investigation.
Ground and prioritize. Wave turns meaningful patterns into opportunities, checking its findings against supporting evidence and prioritizing opportunities by expected impact. Review the evidence, target metrics, and recommended change to decide what deserves attention.
Act. Review and refine an opportunity, route it to Jira, Linear, or GitHub, or hand it to a coding agent. You can also use MCP tools to manage opportunities natively in your IDE, custom harnesses, or desktop agent apps.
Measure and tune. Wave evaluates shipped changes against your target metrics and can use feature flags to control rollout and experiments to test a change's impact. It then uses the results to inform subsequent investigations. You can also provide feedback to help Wave make more relevant recommendations that align closely with your goals.

Key concepts

These concepts explain how Wave organizes its work and how you can guide its recommendations.

Product area

A product area is a focus area that tells Wave where to look and how to judge what matters. It might represent a surface like "Chart Builder", a journey like "Checkout", or a customer outcome like "Successful onboarding".

Each product area captures scope, goals, target metrics, sources, code repositories, and durable guidance. Everything the system discovers and scores happens within that context. For details, go to Product Areas.

Opportunity

An opportunity is a recommended product improvement: a problem to fix, a workflow to improve, a bug to resolve, or a new idea worth testing. Each opportunity includes:

A clear problem statement and proposed solution
Supporting evidence from multiple sources
A priority score and expected impact on target metrics
A lifecycle stage: Triage, In progress, or Shipped
An implementation plan and acceptance criteria for handoff

For details, go to Opportunities.

Evidence

Evidence is what separates an opportunity from a guess. Amplitude cites the charts, dashboards, replays, feedback themes, or other artifacts that support a recommendation, so reviewers can validate the claim before they act.

The opportunity graph

Opportunities aren't a flat backlog. They live in an opportunity graph that connects work to evidence, metrics, pull requests, agents, and the screens or flows in your product map. The graph helps Amplitude reason about relationships between problems, plans, and outcomes. For details, go to The opportunity graph.

Insight sources

Insight sources are the data streams Wave uses to investigate and find product improvements. They include Product Analytics, Session Replay, AI Feedback, experiments, specialized agent results, and code context. The more data Wave has that's relevant to your goals, the more specific and useful its recommendations can be.

For details, go to Connect context.

Feedback

Product area settings are the primary steering mechanism. Ratings, comments, and triage decisions all teach Amplitude what your team values for that area. Over time, that signal helps Wave produce opportunities that better match the product area's scope, goals, and preferred work style.

For details, go to Tune Wave discovery.

What to expect

When you open Wave from the left sidebar, it opens to Chat. You can then navigate to Product Areas, Opportunities, and Settings. You can:

Create an initial set of product areas.
Generate and view opportunities.
Connect the workflows you want to use for execution.
Connect the Wave MCP tools in your preferred AI client.
Review an initial batch of opportunities and give feedback.
Take action on at least one high-confidence item.

Start with review and validation

The easiest way to get started with Wave is to create a product area and review the opportunities it generates. Validate the evidence behind an opportunity, then take one small action like creating a Linear ticket.

How discovery works

Wave starts discovery when you create a product area, when a scheduled investigation runs, or when you choose Create > Generate opportunities or Create opportunity. During an investigation, Wave:

Reviews product data, session replay insights, customer feedback, experiments, and other selected sources.
Groups related findings and investigates them in more depth.
Looks for corroboration across independent signals before promoting a finding.
Scores and prioritizes the results, then creates opportunities with evidence, expected impact, and a draft plan.

The result is a prioritized set of opportunities you can review.

Use Wave with other Amplitude AI tools

Wave uses context and capabilities from across Amplitude AI tools:

Specialized Agents contribute their findings to Wave investigations as one type of insight source.
Global Agent helps you ask follow-up questions, run additional analysis, and refine an opportunity.
The Amplitude MCP server lets you review opportunities and work with coding agents in tools like Claude Code, Cursor, and Codex.

Get started

For ongoing configuration, manage product area schedules, ownership, and limits; tune settings and feedback; and review memories and skills. Connect Jira, Linear, or Slack to bring Wave into your team's workflow.

Was this helpful?