Wingify review: features, pricing, and AI capabilities

Platform overview

Wingz is Wingify's embedded AI layer. It replaces VWO Copilot and AB Tasty's Evi. It ships nine built-in agents (Analytics, Hypothesis, Synthetic Testing, Experiment, Personalization, Rollout, Merchandising, Messaging, Guardrail) that work inside approval gates and kill switches. It turns funnels, replays, forms and surveys into ranked hypotheses, builds variations from a text prompt, Figma file or image, segments visitors for personalization, and runs configuration checks before launch.

Wingify runs two MCP surfaces. The vendor-hosted server at https://mcp.wingify.ai/mcp (with EU and Asia endpoints, OAuth 2.0 with Dynamic Client Registration) reads campaign reports, daily data, variation code, account config, Data Platform segments, heatmaps and session recordings, and can create A/B and Split URL tests from natural language, clone campaigns, update configuration, and start, pause or stop campaigns. The self-hosted FME server (npx wingify-fme-mcp@latest, account ID plus API key) handles feature flags, rollout and test rules, and environment toggles from the IDE.

AB Tasty's older Flagship MCP (@abtasty/mcp-server) is still on GitHub but is not referenced in Wingify's docs.