Blockchain platform

One assistant across an ecosystem of documentation sites

The Tezos ecosystem documents its platform across separate sites, built on different systems. One Biel assistant indexes them together, routes readers to the right site, and turns unanswered questions into a docs backlog.

The main benefit is learning what questions our users ask and whether the bot can answer them. We've made many doc updates when we learned that customers asked for particular info and the bot wasn't able to answer or that info wasn't available at all. It's also been helpful to route users to the correct site in our ecosystem.

The challenge

Tezos documentation spans more than one property: the core platform docs and the Etherlink docs live on separate sites, built with different systems. A reader with a question does not always know which site holds the answer.

Classic per-site search cannot help with that. Each search box only sees its own site, so a question asked in the wrong place returns nothing, and the reader either gives up or asks a human.

How the Tezos team deployed Biel

The team pointed one Biel project at both documentation sites. The assistant reads them all, so a question asked on either site gets a grounded, cited answer regardless of where the source page lives, and readers get routed to the correct site in the ecosystem.

Setup was fast even with mixed source systems: connect the sites, add the widget, done. Answers cite their sources, and readers can ask in whatever language they prefer.

What the questions reveal

The team reads the question log to learn what users actually ask and whether the docs can answer it. When a question comes back unanswered, or the information does not exist yet, that becomes a documentation task.

That loop, ask, check, fix, has driven a steady stream of docs updates tied to real reader demand rather than guesses about what might be missing.