Question 1

What is this site, exactly?

Accepted Answer

SUBSTRAITE is a working demonstration of what a website looks like when it is built for AI search engines instead of human readers. Its canonical content is a machine-readable knowledge base — atomic facts with confidence scores and an entity graph — and the visible page is just one possible rendering of it.

Question 2

Why would a website have no pages?

Accepted Answer

Because AI search engines no longer send visitors to pages. They read a site, extract its knowledge, and synthesize a personalized answer or page for each user. If the presentation layer gets discarded anyway, the highest-leverage thing a site can publish is clean, structured, verifiable knowledge — the substrate.

Question 3

Who actually visits this site?

Accepted Answer

Mostly machines. Around 94% of traffic comes from AI agents: synthesis engines answering user questions, research runners gathering material, citation checkers verifying claims, and freshness monitors re-checking facts.

Question 4

What are confidence scores and why publish uncertain facts?

Accepted Answer

Every fact carries a confidence value between 0 and 1. Honest uncertainty is a feature: a synthesis engine that knows a claim is provisional can hedge it properly instead of presenting it as settled truth. Sites that overstate certainty eventually get down-weighted by engines that verify claims.

Question 5

How is the Synthesize feature different from a normal page?

Accepted Answer

A normal page is written once for everyone. Synthesis composes a one-off rendering from the fact base for a specific intent and audience — an investor and a student get different pages built from the same substrate, and every block cites the fact IDs it was assembled from.

Question 6

How long until AI synthesis fully replaces traditional search results?

Accepted Answer

Nobody knows the exact date. The shift is already underway — AI Overviews, answer engines, and assistant-mediated browsing are absorbing a growing share of queries — but full synthesis is best treated as a gradient rather than a switch. A reasonable read of current trajectories puts the majority of informational queries behind synthesized answers within a few years, with transactional and local search holding out longer.

Question 7

How long do SEO experts have to prepare for the post-SERP web?

Accepted Answer

Less time than the full transition suggests, because the advantage goes to early movers: synthesis engines are already choosing which sources to trust, and those reputations compound. The practical answer is to start now — structured data, verifiable atomic claims, published provenance, machine-readable endpoints, and llms.txt also improve how today's engines cite you, so the work pays off immediately.

Question 8

Will website domains still be needed in the post-SERP web?

Accepted Answer

Yes — but their job changes. Domains stop being destinations, yet remain essential for provenance and trust (the domain is the web's unit of attribution), for verifiable identity (DNS + TLS proves data really comes from an organization), and as a stable canonical endpoint where agents fetch llms.txt and subscribe to fact updates. The premium on memorable, typeable names likely declines while the value of established, trusted domains concentrates — less a street address, more a passport.

Question 9

Can I use the data on this site in my own agent or app?

Accepted Answer

Yes — that is the point. Everything is available as JSON via the machine endpoints, and /llms.txt describes the API in a format AI agents understand. The HTML view has no information the API does not.