LoreAds • Practical guide
How to integrate sponsored ads into an AI chatbot
An ad integration should be small, secure and easy to remove. Keep the assistant's answer independent, add an explicitly labeled placement and design for the no-ad case first.
1. Define the placement
Choose a location beneath the assistant answer where a Sponsored card will not be confused with the assistant's own words. The user should be able to read the answer normally without an ad blocking controls or interrupting text. Plan layouts for mobile, screen readers and narrow screens.
Useful fields for a placement include the sponsor identity, short headline, supporting text, call to action and safe destination link. Do not invent endorsement language or hide the Sponsored label.
2. Keep authentication on the server
Register your chatbot and obtain its integration key through the publisher account workflow. Store the key in a server-side secret manager, never in browser JavaScript, public source code, mobile app bundles or logs. Rotate a key if exposure is suspected.
Read the current LoreAds developer documentation for the actual request and response contract. Copy the documented schema rather than assuming a generic ad SDK format.
3. Send only a minimal topic signal
Classify the current user request into a short, broad, non-sensitive topic (for example, software, productivity or education), or use another minimal signal supported by the documented API. Do not submit full chat logs, personal identifiers, health details, financial account information, precise location or children's information.
Verify that you have appropriate notices and permissions for the information your application processes. The fact that matching is contextual does not eliminate privacy or security obligations.
4. Handle the full response lifecycle
- Ask the assistant to generate its answer independently.
- Determine whether the current topic is eligible for a commercial placement.
- Call the advertising integration from your backend using the documented key and minimum context.
- If the response contains an eligible sponsored placement, show it in a separate clearly labeled card.
- If there is no ad, a timeout, an error or a rate limit, show the normal assistant answer without a sponsored card.
- Use only the documented impression and click recording behavior, with idempotency and anti-fraud controls where supported.
Do not retry indefinitely or degrade the AI answer when ad delivery is unavailable.
5. Test before any public pilot
- Confirm the card is visibly labeled Sponsored on mobile and desktop.
- Test a relevant topic, an unrelated topic and a sensitive topic that should show no ad.
- Test API timeouts, a revoked key, rate limits and an empty response.
- Check navigation destinations, safe URL handling, analytics and accessibility.
- Measure latency and ensure the assistant answer can render if the advertising service fails.
LoreAds' present live commercial endpoint can return a delivery-disabled response while paid operations remain locked. Use the documented isolated testing environment for integration validation; it does not create real billable impressions or publisher earnings.
6. Apply for a beta integration review
Once the integration works in testing, apply for the LoreAds publisher beta and share the chatbot's use case and integration questions. The beta is for evaluation, not a promise of ad fill, income, payouts or production availability.
Frequently asked questions
What if the ad service is unavailable?
Return the normal chatbot answer without displaying a card. Treat sponsored content as optional, never as a requirement for the assistant to function.
Should API keys be embedded in the frontend?
No. Keep credentials on the backend and use server-side requests. A public key leak risks abuse and unauthorized activity.
Can testing generate revenue now?
No. LoreAds evaluation testing is not real-money paid delivery and does not create payable earnings.