company logo

Help center

Go to Rank Prompt
Powered by
All collectionsAgentUsing the agent

Using the agent

Open the agent, ask it about your brand, hand it work, and use commands, mentions and images to get better answers.

Opening the agent

  • The agent button at the bottom right of every page opens the agent in a side dock. Shortcut: press G, then A.

  • Ask agent in a page header opens the dock and runs that page's skill on what is on your screen.

  • Agent in the sidebar opens it full screen, with your conversations in the side panel. Continue in the side dock moves a conversation to the dock so you can keep browsing the app.

The agent works on the brand selected in the brand switcher. On Agency Plus, choose All brands (or Agency mode in the brand switcher) to ask across your whole portfolio, and This brand to go back.


What to ask

Ask in plain words, the way you would ask a colleague who knows your data.

Visibility

  • "What is my Visibility score this month, and what changed?"

  • "Which AI platforms name us least?"

  • "Which prompts miss our brand, and who gets named instead?"

  • "Run a report on ChatGPT, Claude and Google AI Mode (AI Overviews) for these prompts."

Competitors and sources

  • "How do we compare with HubSpot on Perplexity?"

  • "Which sites does AI cite for our category that never mention us?"

  • "Which Reddit threads should we answer first?"

Content

  • "Write an article for the prompt 'best CRM for small teams'."

  • "Tighten the intro of my latest draft and add a FAQ."

Tasks

  • "What should I work on this week?"

  • "Work on the task about our pricing page."

  • "Create a task to update our G2 profile, high priority."

When you hand it a task, the agent moves it to In progress, does the work (reads your data, runs what is needed, drafts the content) and moves it to Done when the deliverable is ready.


Commands and mentions

Type / in the message box for commands:

Command

What it opens

/help

Commands and how mentions work

/tools

What the agent can do

/memory

What the agent remembers

/settings

API keys and web search

/usage

Context and AI budget usage

/github

Connect a GitHub repo for pull requests

/new

Start a fresh conversation

Type @ to mention a task, report, article, audit, campaign, contact and more. The agent uses the exact item you pick.


Actions that ask first

When the agent is about to spend credits, delete something or write outside Rank Prompt, it shows an Action required card with what it will do and the cost. Click Confirm to run it or Cancel to stop. Everything else (reading data, creating a task, writing a draft) runs without asking.


Images

Attach up to 3 images per message: screenshots, charts, a competitor's page. PNG, JPEG, WebP and GIF work; the app shrinks large images before sending them.


Long conversations

A conversation has a context limit. When it gets close, the chat shows Conversation getting long with the share of context used and a New chat button. Past the limit the conversation ends, so start a new one when you see the warning. What the agent saved to memory carries over.


Managing conversations

The side panel lists your conversations. From there you can Search them, start a New chat, open Skills, and rename, delete or Export as Markdown a conversation. Rate an answer with Good response or Poor response to help improve the agent.


Tips

  • Be specific. "How did our ChatGPT visibility change since the last run?" beats "How are we doing?".

  • Name the feature. "Run a report", "create a task", "draft an article" point the agent at the right tool.

  • Follow up. It keeps the whole conversation, so "and on Gemini?" works.

  • Tell it what to remember. "From now on, always compare with last month" is saved for next time. See Agent memory.


Next steps

  • What is the agent?: what it can do, who can use it and the AI budget

  • Agent memory

  • Agent settings and your own OpenAI key


Questions?

Email [email protected] and I'll help.

Did this answer your question?
😞
😐
😁