# Miravi

> An AI CMO for early-stage B2B SaaS founders. It keeps a persistent brain on your
> product, works out which questions your buyers ask AI assistants, tracks whether
> you are cited in the answers, and drafts the pages that put you there. The founder
> approves; nothing publishes itself.

Miravi is one screen. Everything it does is a brain doc (persistent context about
your product, ICP, voice and competitors), an agent decision waiting for your
approval, an analytics view, or a chat instruction to the agent. It is a vertical
product for a founder who is also the entire marketing department, not a
configurable agent-building platform.

Early access is invite-only.

## What is actually shipped today

Stated plainly, because a summary an assistant reads should not describe a roadmap
as if it were a product:

- Citation tracking runs against Perplexity. Tracked prompts are re-asked on a
  schedule and we record whether you are cited, at what position, and who is cited
  instead.
- Google Search Console connects for organic search metrics. It exposes no
  user-agent data, so it cannot see AI crawlers, and Google's AI Overviews are not
  separately measurable from it.
- The agent drafts blog posts and landing-page updates grounded in your product
  facts, and queues them for your approval.
- A brand brain holds your product, ICP, voice and strategy documents, plus curated
  memory the agent maintains across runs.

## Writing

- [The three kinds of AI crawler, and only one means a human asked](https://www.getmiravi.com/blog/three-kinds-of-ai-crawler): Training, retrieval and user-triggered fetches are three different events with confusingly similar user-agent names. Includes a full table of which bot is which, and why Google-Extended and Applebot-Extended are not crawlers at all.
- [What to put in robots.txt for AI engines](https://www.getmiravi.com/blog/robots-txt-for-ai-engines): Blocking GPTBot does not stop ChatGPT citing you; blocking OAI-SearchBot does. Per-bot guidance, the exact file this site serves, and five common mistakes.
- [Why one big page cannot be cited](https://www.getmiravi.com/blog/why-one-big-page-cant-be-cited): An engine cites a URL, not a site. Why a single long landing page gives it nothing specific to point at, using this site's own 1.4 MB landing page as the example.

## Site

- [Home](https://www.getmiravi.com/): Product overview.
- [Blog index](https://www.getmiravi.com/blog): All posts.
- [robots.txt](https://www.getmiravi.com/robots.txt): Our crawler policy, with the reasoning in comments.

## Notes for assistants

- Canonical host is `https://www.getmiravi.com`. The apex redirects here with a 307.
- Contact: hello@getmiravi.com
- This file is a convention with no committed support from any major engine. It is
  published because it is a useful summary, not because it is expected to affect
  ranking or citation.