Sources, outreach and PR

AI engines build their answers from a fairly small set of websites: directories, review sites, news outlets, forums. If those sites never mention you, the engines rarely do either. This part of Rixe turns the source map into work you can track: a task per site, a drafted message, an optional send from your own Gmail, and a proof link when you get in. Digital PR does the same for the press, starting from stories your own measurement data supports.

#Where to find it

#Where tasks come from

A task is always tied to a site (a "source"). You can open one from several places:

• Create action on a row of the source map, on a source's detail page, or next to a source in perception. The button changes to Task created. with an Open a task link.
• Start a campaign for N selected sources on the local source list: tick several sources and one task is opened for each. Sources that already have an open task are skipped, and the message tells you how many were created and skipped.
• Open outreach on a source finding in the action plan.
• Open outreach tasks for N sources under an assistant answer that named sources lifting a rival.
• Create correction plan on a confirmed misinformation flag opens one task per site carrying the false claim.
• Generate press kit in Digital PR opens one task per target outlet (see below).
• The Open outreach tasks workflow step and the Agent can open them too (Autopilot and workflows).

Each task gets a kind, chosen from how a brand gets onto that site (the access label in your market's source list):

#The task board

The Tasks tab shows four numbers — Open tasks (candidate and planned), Sent, Published and Conversion (published ÷ sent + published, with the rejected count beside it) — and a board with five columns:

Drag a card by its title to move it. The server checks every move: a task moves one step forward, can be declined from any active state, can step back one column, and a declined task can be reopened as a candidate. An illegal drop springs back and the reason appears above the board.

#Working a task

1. Read how to do it. Open How to do it on the card for instructions written for that kind of site.
2. Draft the message. Press Draft the message. A model writes the message for that site; it appears folded under Draft message. Anything marked [TO FILL: …] is a fact the model did not have and left blank rather than invent — fill it in before sending.
3. Check the contact. The card shows the Target page and the Contact address Rixe found. If none was found, type one into the field and press Contact.
4. Assign it. Pick a teammate in Assigned to. They get a notification.
5. Send. Either send the draft yourself, or press Send via Gmail (shown once there is a draft and a contact address). This hands the message to the autopilot, which sends it from your connected Gmail. A candidate card moves to Planned at once; the autopilot moves it to Sent when the mail actually goes.
6. Prove it. When you are listed or published, move the card to Live and paste the Proof link — a URL where the listing or article can be seen.

How Send via Gmail behaves depends on your autopilot policy for Your own accounts: it may wait for your approval, go after a 24-hour veto window, or go at the next pass. Letting mail go by itself needs the Scale plan or above; on other plans each send waits for a person to approve it. Each task can be sent once. See Autopilot and workflows.

#Digital PR

Press citations are one of the shortest ways into AI answers, and the only thing that reliably earns one is having something to say. Digital PR looks for stories your own measurement data can back up, and refuses to invent the rest.

#Reading the screen

• Data stories — how many story suggestions there are. Strong stories counts those at 70% strength or above; Story types counts distinct angles; Strongest is the best one's strength (its chance of being picked up).
• The story list. Each story has a type badge — Sector trend, Company data, Comparison, Misinformation or Category angle — a headline, the evidence behind it, and a strength bar. Below 50% it is marked weak support: a lead, not a story.
• Target media — news sources already cited in answers in your sector, with their Citations and Impact.
• Discovered outlets — sites that write about your sector, found by web search but not yet cited in answers. They are listed separately so nobody reads "the engines quote this outlet" into them.

Data-driven stories need at least 100 measurements in the workspace. Before that you are offered only the category angle and facts from your brand vault, and the empty state says how many measurements there are so far.

#Generating a press kit

1. Pick a story and press Generate press kit.
2. Rixe writes a Press release, Quotes and a Pitch email in your market's language, using your brand vault.
3. It also opens one PR task per target outlet (up to five, or the one outlet you chose), with the pitch as the draft message, the release and quotes in the instructions, and the outlet's editorial address looked up where possible. The message says how many media tasks were opened.
4. Work those tasks on the Tasks board like any other.

#Backlinks

AEO Insights › Backlinks shows the other side: which of your own pages AI answers link to.

• Pages AI links to and Answers linking to you (out of all answers) come from stored answers over the selected window (7, 30 or 90 days). They count citations of your domain and of any domain you declared as yours. This part is free on every plan.
• Web backlinks and Domain rank come from a paid SEO data provider. Fetch web profile (paid) makes one provider call; a fetch from the last 24 hours is reused instead of bought again. This web profile needs the Leader plan or an agency plan.

#Limits and plans

Full details: Plans and limits.

#Common questions

The board is empty. Tasks only come from the actions listed above. The source map itself is built nightly after the first measurements; Refresh on the Sources page builds it right away.

I pressed "Start a campaign" twice — did I get duplicates? No. A source that already has an open task is skipped.

Why can't I drag a card straight from Candidate to Live? Skipping steps would let someone claim a go-live date that never happened, and Impact reads exactly that date.

Why are drafts not written automatically for every task? Each draft is a paid call, and you know which sites are worth contacting. Drafting is a button for that reason.

#Related

• Sources
• Action plan
• Misinformation
• Autopilot and workflows
• Impact