# Ideas for advertising in AI apps.

Campaign strategy, publisher monetization and the experience between them.

- [One URL, two readers: serving the Jolts site as Markdown](https://jolts.xyz/blog/one-url-two-readers.md): What we found when we fetched every public Jolts page three ways, including a bug in our own audit.
- [The sponsored card that arrived one turn late](https://jolts.xyz/blog/the-card-that-arrived-one-turn-late.md): A two-turn test for keeping a late ad response out of the wrong conversation.
- [What happens when a sponsored card says “run this command”?](https://jolts.xyz/blog/the-ad-an-agent-must-not-execute.md): A concrete boundary between paid creative, model context and tool execution in coding assistants.
- [The chat reconnected. Did you count the ad twice?](https://jolts.xyz/blog/one-impression-after-reconnect.md): Walk one sponsored impression through a lost acknowledgement, a retry and a page reload.
- [What can an ad matcher learn from four fields?](https://jolts.xyz/blog/context-without-the-transcript.md): Design a small context payload and test which details actually improve the match.
- [Can one sponsored placement pay for an AI answer?](https://jolts.xyz/blog/can-one-placement-pay-for-an-answer.md): A worked example of the economics of a free tier, with fill, eligibility and publisher share shown separately.
- [Build a test set where the right ad is no ad](https://jolts.xyz/blog/a-test-set-for-showing-nothing.md): Use pairs of nearly identical prompts to find false positives before a publisher pilot.

[Browse articles on the website](https://jolts.xyz/blog)