Ads run beside your AI,
not through it.

The infrastructure between your AI app and the ad market.

Your app talks to one endpoint. wavebird runs the ad path behind it and returns a single decision per placement, so your app never has to speak the ad market’s protocols itself.

See where wavebird fits by use case →

One request. Two independent paths.

Your model generates the answer while wavebird prepares the ad at the same time. Neither path waits for the other, and neither one changes the other.

The ad market never gets the conversation.

wavebird reduces the request to the minimum signal needed for ad matching. Advertisers receive a category and a language, not the original prompt, the user identity, or the chat history.

Not every AI moment should show an ad.

wavebird checks the context before matching an ad. Sensitive or unsuitable moments return no ad at all. You decide which categories and advertisers are allowed inside your product.

The ad is ready while the model is still thinking.

Because the ad path runs in parallel, the placement can be shown during generation. You monetize the wait, not the answer, and not the user’s patience.

Every paid impression has proof.

wavebird tracks each delivered placement from ad decision to impression. Signed evidence creates one clear record for measurement, reporting and payout, the same record you and the advertiser are billed against.

wavebird follows a published compute sponsoring specification that defines separation, data handling, measurement and publisher control. Read Compute Sponsoring v1.0 → See how delivery is measured → See what leaves your app →

This is the request behind the flow.

Your backend asks for an ad decision with the placement, timing and consent settings you choose. wavebird returns the result and your app renders it, inside your own UI.

1const decision = await fetch("https://api.wavebird.ai/v1/placements", {2  method: "POST",3  headers: {4    Authorization: "Bearer " + process.env.WAVEBIRD_KEY,5    "Content-Type": "application/json",6  },7  body: JSON.stringify({8    client_id: "wavebird_demo_app",9    session_id: "sess_4f2a91",10    slot_hint: { position: "below", max_width: 720 },11    overrides: { allowed_formats: ["banner"], timing: "during" },12  }),13});14const { placement } = await decision.json();

Why not just add ads yourself?

Traditional ad infrastructure was built around pages, feeds and video. wavebird is built around the AI response and the generation moment, which is where the inventory actually is.

Ready to put wavebird in your AI app?
Get your API key, connect the first placement, and see how ads behave inside your own product.

No upfront fees, we only earn when you do.