How We Got LLMs to Recommend Our Open Source Library — Christopher Burns, Inth
AI Engineer World's Fair 2026 · 16:27
← All organizations
Privacy engineering and consent infrastructure
Inth
Inth builds privacy engineering tools for developers and compliance teams. Its code audits trace personal-data flows, detect trackers and SDKs, and connect vendor calls with consent and policy records. Teams can review findings alongside pull requests, receive proposed code fixes where possible, and retain decisions and fix histories. Its website audits inspect live scripts, cookies and destinations, comparing production behavior with visitor consent, regional rules and privacy disclosures.
Founded in 2025 and led by founder and CEO Christopher Burns, the company began as Consent.io. It launched its open-source consent library, c15t, in 2025 and adopted the Inth name in 2026 as its scope expanded beyond cookie banners. Its broader direction includes AI agents that help legal, development and marketing teams collaborate around a codebase rather than disconnected policy documents.
c15t’s consent platform separates the interface from backend infrastructure: developers control React components and styling, while Inth’s hosted services resolve regional policies and store consent records. Visitor choices govern scripts, iframes and subscribed application code. This connects a customizable website interface with centrally managed consent behavior and records. In its Y Combinator profile, the company reports more than three million npm downloads for c15t.
Explore the recordings
This Inth archive currently contains one supplied recording: Christopher Burns’s practical account of making c15t documentation easier for LLMs and coding agents to use. The paths below highlight different parts of that talk. Reported results and recommendations describe the recording; they do not establish today’s product capabilities or performance.
Start with the website documentation
For a documentation-site starting point, watch How We Got LLMs to Recommend Our Open Source Library. Burns recommends Markdown versions of pages, a manually curated llms.txt, and a fetchable page index with short descriptions. He explains several ways to serve Markdown and describes early WebMCP tools for searching and retrieving documentation. His Q&A emphasizes Markdown pages where feasible and calls llms.txt the ‘first shout’, with the fuller index second.
Follow the package into the coding agent’s workspace
Return to Burns’s talk for the package-distribution path: bundled Markdown documentation and an AGENTS.md file intended to help agents inspecting repositories or node_modules. Burns reports nearly 50% token savings across multiple models compared with finding documentation on the web. Treat that figure as his reported test result, rather than a general guarantee.
Read the referral results alongside the evaluation caveats
Use the same recording to examine how Burns connects documentation changes to discovery. He reports an onboarding referral spike beginning April 13 and says Claude, ChatGPT, Codex, and Gemini collectively became c15t’s leading inbound source. He attributes the outcome to many small changes rather than one optimization. His discussion of evaluation tooling also warns that scores and recommended practices change rapidly; this archive does not independently verify causation or current referral performance.
1 talk
Newest first1 speaker at AIE
Affiliations reflect their AIE appearances, not necessarily current employment.