Ready for AI agents

Make the information on your website easier for search engines and
AI tools to access and interpret. We review crawl access, page
content, structured data and the interfaces your application
exposes, then implement the agreed changes. This site provides an
example you can inspect. Inclusion in search results or an AI answer
remains the provider's decision.

Discovery settings and indexing signals

We review crawler access through robots.txt, the sitemap and
search-indexing settings. This site's Content Signals policy permits
search, AI answers and training; your policy may differ. Pages
intended for indexing belong in the sitemap with truthful
modification dates. IndexNow announces changes to participating
engines; receiving a notification does not guarantee that a page is
crawled, indexed or ranked.

Read: the text, not the furniture

A page built for eyes is mostly navigation, script and layout, and
an agent that pays for every word it reads pays for all of it to
reach the paragraph it came for. Every page here has a Markdown twin
at the same address: add `index.md`, or ask for `text/markdown` and
the server answers in kind. `llms.txt` is a map of the whole site
written for a language model, `llms-full.txt` is the site in one
file, and every page carries structured data that names the company,
its services and where the page sits among the others. Every home
page answers with a `Link` header, `rel="describedby"`, that points
at `llms.txt`, so an agent learns the map exists before it reads a
word of the page.

Usable: able to act, not only to read

Many agents that work a browser read a page the way a screen reader
does, through its accessibility tree. The work that makes a site
usable without a mouse — real buttons, labelled forms, headings in
order, landmarks — is the same work that lets an agent find the form
and fill it in. This site answers every criterion of WCAG 2.2 AA in
writing. For an application, usable means a described API: an
OpenAPI document, an MCP server where agents should call it
directly, and the OAuth metadata that lets an agent act for a user
without a password changing hands.

A decision on every recommendation

New proposals for agents arrive faster than they settle, and a
scanner will report every one of them missing. We classify each
finding rather than obey it: what it does, who reads it, and what it
costs to keep true. On this site that meant declining five. One
robots rule for every agent rather than one per agent, because a
rule written for one crawler silently replaces the general one for
it. No API catalogue and no OAuth metadata, because there is no API.
No agent card and no commerce file, because there is nothing here
for an agent to call or buy. A client gets the same thing: a list of
what to add, and a list of what to leave alone, each line with its
reason.

Check it from your own browser

Nothing on this page has to be taken on trust. Read robots.txt,
llms.txt, agents.txt and ai.txt, or open this page as Markdown. The
build reads every twin for leftover markup and every sitemap date
against the page's history, and it refuses to publish when either is
wrong — so the half of this site that agents read cannot quietly
fall behind the half that people do.

How an engagement runs

It starts read-only. We look at what agents get from you today — the
robots file, the sitemap, the text behind the layout, the structured
data, the accessibility tree and, for an application, the API — and
hand you the findings, each with a decision and its reason. The
changes are then made in your repository, through your review, with
checks that fail the build when a page stops being readable, so the
work does not decay once we leave. It is bought like the rest of our
work, on the terms in what it costs.

If you want agents to find you, read you right and get things done
for the people who sent them, tell us what you run (read by the
engineer).

hThis page on the web	URL:https://engineer.company/ai-ready/	engineer.company	70
