For the complete documentation index, see llms.txt. Every page on this site is also served as Markdown: append `.md` to any URL, or send `Accept: text/markdown`.
Skip to content
Attensira LogoAttensira
Book a demo

Standing work the crew does without being asked

An automation is an instruction the crew carries out on a schedule or when something changes: a competitor edits a page, a citation is lost, a prompt slips. Every workspace starts with the Daily win plan, up to three pieces of work each morning. Each run is stored, says why it started, and spends credits only when it runs.

Start freeBook a demo

7 days free · card required

Automationsattensira.com · 2 active
Chief of staff agentEvery day · 06:00 UTC
Daily win plan

Load the house skill orchestrate/daily-win-plan and follow it. Last ran 4 Oct, 06:01. Next run 5 Oct, 06:00.

githubrewriteoutreachcommunity
Analyst agentOn a change
Event responder

For each event, say what changed and which tracked prompts it puts at risk, then take at most one next action. If an event threatens no tracked prompt, say so in one line and do nothing.

prompt_lostcitation_lostcompetitor_changednew_deciding_sourcepage_changed

Inbox · “Daily win plan did not finish”: a failed 3am run lands where someone will see it, not in a log.

The instructions are quoted word for word.

Chief of staff agentChief of staffattensira.com workspace, read over the Attensira MCP ·

  • Chief of staff agentChief of staff · Runs the Daily win plan each morning and hands each piece of work to the role that does it.

  • Analyst agentAnalyst · Reads what changed since the last run, and says when a movement is too small to call.

How does an automation run?

It is the same request you would type into a chat, attached to a trigger so you stop having to ask.

  1. Pick a trigger

    Every day, every week on a day you choose, on a change, or only when you ask. Client reports add monthly and quarterly. Times are in the automation's own time zone.

  2. Write the task, not the topic

    “Weekly visibility” gives the crew nothing to do. “Compare mention rate for the Pricing topic with the previous 7 days and list what cleared the 95% test” does.

  3. Each run says why

    Every run starts with one sentence built from recorded facts: Because “Weekly client report” runs every Monday at 08:00, or Because a competitor changed pages on its site.

  4. Work still waits for you

    Anything a run would publish, send or spend follows your approval mode. An approval left for 7 days closes the run having changed nothing.

What happens when an automation run fails?

It does not fail quietly. A run that stops lands in the Inbox with what happened and how to start it again. Both rows below came from scheduled runs on our own workspace.

Inbox2 waiting for you
  • Chief of staff agentDoDaily win plan did not finish

    Run failed · since 23 Sep

    The run stopped before it finished. It has happened 7 times; open the automation to run it again.

  • Chief of staff agentDoDaily check-in did not finish

    Run failed · since 15 Sep

    This run was queued but nothing ever picked it up, so it was stopped. Ask again and it will start fresh.

attensira.com Inbox (role shown is the one whose tool makes the item), read over the Attensira MCP ·

Which automations come ready to use?

Five starters open the same dialog pre-filled. A starter you already run shows Running or Paused and links to yours.

Which automations come ready to use?
Daily win plan · every day, 06:00Up to three pieces of work a day toward the prompts you want to win. Every workspace has it from the start.
Event responder · on a changeWakes when a prompt is lost, a citation drops, a competitor edits a page or a new deciding source appears, and takes at most one action per event.
Weekly client report · Monday 08:00A report built from counted facts, each with its sample size. Monthly and quarterly versions come with the client-report templates.
Reddit watch · Tuesday 09:00A weekly starter for Reddit. Community replies are always drafts for you to post from your own account; Attensira never posts to a community.
Content refresh · Wednesday 07:00A weekly starter for updating your existing pages. Any change it proposes lands as a pull request, a CMS draft or new wording, and waits under your approval mode.
Your ownAny instruction you would type in a chat. The crew can also offer to save a recurring request as an automation; nothing is created until you confirm.

What can an automation not do?

It is a schedule around the crew, so it inherits every limit the crew has, and adds two of its own.

  • It cannot make data fresher than the reading. Each prompt, engine and country is read at most once a day, so an hourly automation re-reads the same answers and spends credits doing it.
  • It cannot report a movement it cannot prove. A change below the 95% significance test comes back as no change, not a small number with an arrow.
  • It cannot run for free. Creating one costs nothing; every run, scheduled, manual or test, is paid for like any other work.
  • It cannot skip an approval. Publishing, sending and spending wait for you exactly as they would in a chat.
  • It is not a workflow builder with branches. It is one instruction, on one trigger, carried out by the crew.

What does a scheduled read bring in?

The Daily win plan starts each morning from answers stored like this one: Perplexity on 4 Oct, defining generative engine optimization from Wikipedia, Contentful and Built In. It plans from what was read, not from a guess.

Answer · Perplexityasked

Question: what is generative engine optimization

Perplexity answered: Generative Engine Optimization (GEO) is the practice of structuring and optimizing content so AI systems cite, summarize, or include it in their generated answers, rather than just ranking for traditional search results. It focuses on being a trusted source that AI tools draw from when answering user queries.

Sources cited

  1. 1en.wikipedia.org
  2. 2contentful.com
  3. 3builtin.com
  4. 4dictionary.cambridge.org
  5. 5coursera.org
  6. 6yotpo.com
  7. +4 more

Not namedAttensira is missing from this Perplexity answer.

Researcher agentResearcher · Stored the answerattensira.com stored Perplexity answer, read over the Attensira MCP · · 10 links

What else do people ask about automations?

How often should an automation run?
No more often than the data changes. Each prompt, engine and country is read at most once a day per workspace, so weekly is a good default for reporting and daily for a watch on something moving. Event automations run only when something actually happens.
What happens when a scheduled run fails or is skipped?
A failed run is raised in your Inbox, and a repeat updates the same row with a count instead of adding one a night. A run that did not happen, because credits ran out or a project is paused, shows in loop health with the reason.
What happens when the credits run out?
Scheduled automations stop, and nothing is charged for runs that did not start. You get one Inbox row and one email until the balance is topped up. A monthly credit cap pauses scheduled runs the same way.
What if my script retries a run?
Send an idempotency key. Over MCP, run_automation takes idempotency_key; over the REST API you send the Idempotency-Key header. The same key resolves to the first run rather than starting and billing a second one.
Can I try an automation before switching it on?
Yes. Test runs it once, now, even while it is switched off. A test is a real, paid run under the same approval mode, but it does not count as the week's scheduled run.
Can I manage automations from my own tools?
Yes. The MCP server has list, get, create, update, run and delete, and the REST API mirrors them. Only run_automation spends credits. No tool takes a workspace id: the key you sign in with names the workspace.

Set the work once. The crew does it every morning.

Start freeBook a demo

7 days free · card required