Inferencefeature
Inference: every model that matters, one POST away
inference.wave.online is live: one standard endpoint fronting 13 providers with measured routing, automatic failover, and per-token metering. Spend is tracked to eight decimals. The public leaderboard shows what we measured and what actually ran. The wave CLI is on npm. Every WAVE surface now serves both people and agents.
Platformfeature
Every WAVE surface is agent-ready: 22 surfaces with full agent discovery
Every product subdomain now serves both people and agents: llms.txt with when-to-use guidance, /.well-known/skills.json discovery, OpenAPI with scoped permissions and typed error models, agent-friendly markdown 404s, and RFC RateLimit headers. The CLI is on npm. The platform usage endpoint shows all 7 channels live.