The crew that does your SEO work
AI agents do. Each project gets a crew staffed from 11 roles, led by a Chief of staff that answers you, hands out the work and reports what changed. No person on our side does client work. You set the direction and approve what ships: nothing irreversible happens without your yes.
7 days free · card required
Question: how to block AI crawlers in robots.txt
Perplexity answered: In the site’s root robots.txt, add a separate block for each crawler you want to block, listing its User-agent and a Disallow: / (or specific paths you want hidden). … Note: robots.txt is a polite instruction, not a secure access control; blocked URLs can still be discovered and accessed.
Sources cited
- 1scalableloops.ai
- 2qwairy.co
- 3flaviocopes.com
- 4maxaeo.ai
- 5geotoolbox.ai
- 6tooltrace.io
- +3 more
Not namedAttensira is missing from this Perplexity answer.
Researcher · Read it and logged the lossStored answer, read over the Attensira MCP, excerpt · Answer “how do I block AI crawlers” at the top of the robots.txt guide
attensira2 checks passed
guides/robots-txt.md
- unchanged: ## Blocking AI crawlers
- added: Add one User-agent group per crawler with Disallow: /. Blocking GPTBot stops training fetches; it does not stop ChatGPT search, which fetches as OAI-SearchBot.
Web engineer · Opened the pull requestExample · example workspaceWho is on the crew?
11 roles. The Chief of staff staffs each project from them and wakes each one when something it is responsible for changes. You never define an agent.
Leads the project
The one you talk to. It staffs the crew, hands out the work and reports what changed.

Chief of staff
Leads the project: answers you, hands work to the crew, makes sure it gets done and reports what changed. Does not do the work itself.
Wakes when You post in the project, and every week to report what the crew did and what moved. Always on the crew.
Works with: your tracked data · hand-offs to projects · Make this a project · the weekly report
Docs: Delegation
Finds what decides the answers
Reads who the answers name, which pages they lean on, and whether a change moved anything.
Researcher · Deciding sourcessince Google's own documentation decides 7 of our losses
- “how to block AI crawlers in robots.txt”6 of 6
- “how to see which AI bots crawl my website”6 of 6
- “alternative to hiring a GEO agency”6 of 6
Tried Read the deciding-source map for 29 to 30 Sep 2026: 330 answers, 8,319 citations.

Researcher
Watches competitors and the sources AI answers cite, and writes up what changed and why.
Wakes when A competitor takes one of your questions, or a new source starts deciding the answers.
Works with: deciding sources · web_search · web_fetch
Docs: Who decides an answer
Analyst
Reads what moved each week, with its sample size, and says whether it is a change or noise.
Wakes when Every week, to read what moved and whether it is more than noise.
Works with: your readings · re-reads at +7, +14 and +30 days
Docs: Before and after
Makes the change on your site
Articles, rewritten passages, images and fixes, filed as drafts and pull requests for your yes.
Writer · New articlesince AI crawler logs analysis tool: what to use, and what the log actually tells you
Needs you Ready for your yes. Articles wait for a person in every mode.

Writer
Drafts new pages and refreshes old ones so they answer the questions you are losing.
Wakes when A question you track has no page that answers it, or one of your pages stops being cited.
Works with: write_document · propose_fix · propose_blog_update
Docs: Articles and posts
Copywriter
Rewrites the passages AI answers get wrong: pricing, product claims and positioning, in your voice.
Wakes when AI answers get your pricing, product or positioning wrong.
Works with: propose_edit
Docs: Rewrites
Designer
Makes the images, charts and social cards that go with the crew's pages and posts.
Wakes when A page AI answers already cite could travel further with a picture or a card.
Works with: generate_image · Canva (read only)
Docs: The agent's tools
Web engineer
Keeps the site readable to AI crawlers and search engines: robots rules, structured data, redirects, sitemaps and rendering.
Wakes when Something on your site stops AI crawlers or search engines from reading it.
Works with: GitHub · resolve_url_to_file · propose_fix
Docs: GitHub
Reaches the places answers read
Pitches, community replies, video metadata and ads: the sources off your site that decide answers.
Outreach · Decisionsince Choose: benchmark or pitch for llmpulse.ai and seranking.com
Tried Both added as competitors. It is not appropriate to draft to the rivals' own inboxes.
Needs you Decide whether this week's target is the third-party roundups our answers cite, or a profile that earns a mention in the next one.

Outreach
Finds the sites AI answers cite, works out who runs them, and drafts the pitch for you to send.
Wakes when A site AI answers cite could mention you, or a pitch is due a follow-up.
Works with: find_outreach_targets · find_contact · draft_outreach · propose_listing
Docs: Off-site work
Community
Finds the Reddit, Hacker News, Stack Overflow and Quora threads AI answers lean on and drafts honest replies for you to post yourself.
Wakes when A thread AI answers cite asks something you can answer.
Works with: find_community_threads · propose_community_answer
Docs: Community answers
Video
Works on your videos: the titles, descriptions, chapters and links that AI answers read.
Wakes when A video AI answers cite needs a better title, description or chapters.
Works with: YouTube (read only)
Docs: Connectors
Ads strategist
Finds paid spend on questions AI answers already give away, and ads that contradict what the site says.
Wakes when You pay for clicks on a question AI answers already give away, or an ad contradicts your pricing.
Works with: Google Ads keyword research · Search campaign drafts
Docs: Google Ads
Tools are the ones the docs name for that kind of work. Each automation's tool groups decide exactly what a run may reach, and a crew member whose integration is not connected works without it. The role shown on an Inbox item is the one whose tool makes that kind of item; the Inbox does not record which role made it.
How does work reach the crew?
You write what you want, one piece of work per line. The Chief of staff hands each line to the project whose crew does that kind of work.
- 01
You ask
In your own words, on Home, in a chat or from your own assistant over MCP. Up to 20 lines in one hand-off.
- 02
It routes
A line that names a #project goes there; with only one project taking work, it goes to that one. Otherwise Attensira picks the project when it is at least 60% sure, and asks you when it is not.
- 03
The crew works
Each line becomes a task in its project, gated again by the stricter of the workspace's approval mode and the project's.
What is left for you?
Steering and saying yes. The agents stop only when they need you, and each stop lands in your Inbox as one of three kinds of row.
Approve
A yes or no on something the agents made: a change to your site, an article, an action that cannot be undone.
Approve it, edit a draft and then approve it, or decline it and say why.
Answer
A fact only you have: a question an agent asked, or a gap in your business profile.
Answer in the row.
Do
Something only you can do: connect an integration, send a drafted email from your mailbox, paste an article where it goes.
Do it, then mark it done.
Nothing irreversible without your yes
- New workspaces start in Draft: nothing is published, sent or written outside the workspace until you approve it.
- Anything that states a claim (a price, a number, a product or competitor name) waits for your yes in every mode.
- We open pull requests; you merge them. Auto-merge, if you switch it on, covers only safe changes such as structured data, meta tags, alt text and internal links.
- A project can be held stricter than its workspace, never looser.
Researcher · Decisionsince OpenAI's help centre decides 6 of our losses
Tried On ChatGPT, help.openai.com is the leading deciding source for six of our questions, cited in 6 of 6 answers each.
Needs you Your decision: treat these as lost, or write the page OpenAI's help centre leaves open.
Analyst · Blockedsince Install the edge SDK so we can see whether AI crawlers reach the site
Tried Read the AI traffic report for 30 days. No edge collector has ever reported.
Needs you Install the edge worker, so we can tell “read and not cited” from “never fetched”.
Questions about the crew
Who does the work at Attensira?
Is a person doing my work behind the scenes?
Can I build or pick my own agents?
What can the crew do without asking me?
Will an agent post or send anything as me?
What does the crew cost?
Put the crew to work on your questions.
Give it your domain. The agents read your site, propose the projects and their crews, and the first work lands in your Inbox for a yes.
7 days free · card required