For technical writers and docs teams

Find out what your docs cannot answer

Biel answers reader questions from your documentation, with sources cited on every response. In return, you get the thing writers never had: a log of every question, every gap, and every rating, turned into a ranked backlog of what to write next.

14-day freeSelf-serveLive in 15 min
Biel.ai chatbot
Powered by biel.ai
Assistant
Hi! how can I help you?
Suggested questions
enable dark mode
Assistant

To enable dark mode for Biel.ai on your website, follow these steps:

  1. Modify your website’s main HTML file.
  2. Add the attribute data-theme="dark" to the <html> tag.
html
<html data-theme="dark">
  <!-- Your existing HTML content -->
</html>

This change activates the dark mode styling for Biel.ai…

Send
AI-generated answers may contain errors. Verify official sources before use.

100,000+ technical questions answered, 98.6% rated helpful

ScyllaDB
Katalon
Tezos
Talon.one
GrepTime
CrazyGames
SecuroSys
ScyllaDB
Katalon
Tezos
Talon.one
GrepTime
CrazyGames
SecuroSys
Why docs teams run Biel

Your docs get more usable. You get the data.

Readers stop hunting through pages and start getting cited answers. You stop guessing what is missing and start working from a demand-ranked list.

i.

A ranked backlog of what to write next

Every question the assistant cannot answer from your docs is logged and clustered by topic. That is your content-gap report: a prioritized list of missing pages, driven by real reader demand instead of guesses about what might be useful.

ii.

Reader questions, verbatim

You finally see how readers actually phrase things. When users ask "why am I getting a 401" and your page is titled "Authentication lifecycle", that mismatch is your next retitle. The question log is continuous, free user research.

iii.

Numbers that prove the docs are working

Satisfaction ratings, resolution rates, and question volume give you evidence for the review cycle. Docs work is notoriously hard to measure; a thumbs-up rate on answers drawn from your pages is a number leadership understands.

iv.

Answers that stay inside your docs

Responses are generated only from your indexed content and cite the source page. No invented flags, no answers from someone else’s product. When the docs do not cover it, the assistant says so, and that gap lands in your report.

v.

Every site you maintain, one assistant

Guide site, API reference, wiki, even sites on different platforms: one index covers them all, and readers get routed to the right site. The Tezos team runs one assistant across docs.tezos.com and docs.etherlink.com.

vi.

Installs without an engineering ticket

One script tag, or a native plugin for Docusaurus, MkDocs, and Sphinx. Most writers have it live in about 15 minutes, styled to match the site. Self-serve, from $50/month.

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.

Tim McMakin · Technical Writer at Tezos ecosystem

What docs teams ask before starting

Will it invent answers our docs do not contain?

No. Answers are grounded in your indexed documentation with source links on every response. When the content does not exist, the assistant says it cannot find the answer, and the question is logged as a content gap for you.

How do I see what readers asked?

The analytics dashboard shows every question, its answer, the rating it received, and the unanswered questions clustered by topic. We describe the full workflow in how technical writers use chatbot analytics.

Can it cover several docs sites at once?

Yes. One project can index multiple sites on different platforms into a single assistant, which also routes readers to the right site. See how the Tezos team runs it across their ecosystem.

Do I need engineering help to install it?

Usually not. It is one script tag or a native plugin for Docusaurus, MkDocs, and Sphinx, plus appearance settings in the dashboard.

How do we measure whether it is working?

Watch satisfaction, unanswered rate, and ticket volume. We published the honest measuring method, including what healthy numbers look like, in how to tell if your documentation chatbot is actually working.

We already have page-view analytics. What does this add?

Page views tell you what a reader opened, not whether their question got answered. The question log shows intent in the reader’s own words, and the unanswered cluster shows exactly where the docs fell short.

Does it replace our existing docs search?

No. It layers on top of your navigation and search rather than replacing them. Readers who want a specific page keep using search, and the assistant handles the questions that span several pages.

What does it cost?

Self-serve plans from $50/month with public pricing, and a 14-day free trial on the Professional plan, no credit card required. See pricing.

Stop guessing what your readers need

14-day free trial · No credit card required
Try me ↓