Frequently asked questions

Install the JavaScript Publisher SDK via npm or CDN script tag. Initialize it with your API key. Call requestAd() when your app receives a user prompt — the auction runs in parallel while your AI generates its response. When the response is ready, call renderAd() to display the winning suggestion. Full docs at /docs.

Revenue depends on your conversation volume, user intent quality, and the categories your app covers. Builders receive 60–75% of ad revenue generated from their app. High-intent verticals like travel, SaaS, and developer tools tend to command higher CPMs. Your dashboard shows real-time earnings so you can see exactly what each conversation is worth.

That's the whole point of PromptBid. Sponsored suggestions are contextually relevant text that appears alongside the AI response — not banners, not popups, not interstitials. Users see a useful recommendation, not an ad interruption. You control the placement, styling, and categories. If a suggestion doesn't match the conversation, it doesn't show.

Yes. You set category blocklists to exclude verticals you don't want (gambling, adult, competitors, etc.). You can also set frequency caps and control how often suggestions appear per session. The builder always has final say over what shows up in their app.

Any AI application with a conversational interface — chatbots, coding assistants, writing tools, research platforms, customer support agents, tutoring apps. If your app takes a user prompt and generates a response, PromptBid can run an auction on that interaction.

Earnings accrue daily and are visible in real-time on your dashboard. Payouts are processed via Stripe on a regular schedule. You can track revenue by day, placement, and category to understand which conversations are most valuable.

AI conversations carry richer intent signals than search queries or page views. When someone asks an AI assistant about hotel options in Tokyo, that's a clear buying signal — stronger than a keyword match, more immediate than a retargeting cookie. PromptBid lets you reach users at the exact moment they're asking for something relevant to your product.

All targeting is contextual. You bid on conversation categories and topics — not user profiles. When a conversation matches your targeting criteria, your bid enters the real-time auction. If you win, your sponsored suggestion appears alongside the AI response. No personal data is used. No cookies. No cross-app tracking.

A sponsored suggestion — a short text recommendation that appears naturally inside the AI response. It includes your brand name, a brief message, and a link. It looks like a helpful suggestion, not a banner ad. You can see a live example on our sandbox page.

PromptBid uses CPM (cost per thousand impressions) pricing. You set your maximum bid and daily budget. The auction determines the actual price — you only pay what's needed to win the impression. No minimum spend required, and you can pause campaigns at any time.

Your dashboard shows impressions, clicks, CTR, spend, and CPM across campaigns. You can segment by category, time period, and placement. All data updates in real-time so you can optimize campaigns as they run.

PromptBid implements OpenRTB 2.6, the industry standard for real-time bidding. This means demand-side platforms and advertisers can integrate using familiar tools and bid request/response formats. We've added privacy-first extensions that exclude personal data from the bid stream.

When your app calls requestAd(), the SDK sends conversation context (categories, keywords — never raw text) to PromptBid's auction service. Eligible advertisers bid in real-time. The highest bidder wins. The SDK returns the winning creative, which you render with renderAd(). The entire auction completes while your AI is still generating its response, so there's no added latency for the user.

Only conversation metadata — categories and keywords that you configure in the SDK. PromptBid never sees the raw user prompt, the AI response, or any personally identifiable information. You control the context mapping. The SDK documentation covers exactly what's sent and how to customize it.

No. The auction runs in parallel with your AI's response generation. Since most AI responses take hundreds of milliseconds to seconds, the auction completes well before the response is ready. If the auction doesn't return in time, the SDK gracefully skips — your app's response is never delayed.

The Publisher SDK and client libraries are open source on GitHub under a permissive license. You can inspect exactly what the SDK does, how it communicates with the auction service, and what data it sends. The core auction infrastructure is proprietary.