Your First Chat
Learn how to talk to the agent, what kinds of questions work best, and how it uses tools to work from your real search data before answering.
How the Agent Works
When you send a message, you're not just prompting a language model — you're talking to an autonomous agent that uses tools. The agent reads your message, decides which tools to call (GSC queries, site analysis, brief generation), executes them, and synthesizes everything into an answer.
The agent can chain up to 10 tool calls per message, building a complete picture before responding. It maintains a 20-message conversation history, so it remembers what you discussed earlier in the session.
Unlike generic AI chatbots, the agent is built to pull your actual data before making any recommendation. It won't guess — it works from your real GSC numbers and crawled page content, not a blank prompt.
What You See in Chat
As the agent works, you'll see its thinking process in real time:
- Tool calls — the agent shows which tools it's using (e.g., “Querying GSC for top keywords...”)
- Tool results — raw data from tool executions appears inline
- Analysis — the agent's final response with insights, tables, and recommendations
This transparency lets you see exactly what data the agent based its advice on — no black-box answers.
Great First Questions
With GSC Connected
If you've connected Google Search Console and synced your data, start with these:
“Give me an SEO overview of my site — top keywords, best pages, and any issues you spot”
The agent runs a broad GSC query and gives you a snapshot of your organic search health.
“Which of my pages have the most impressions but lowest CTR? Suggest improvements.”
Finds pages where people see your listing but don't click — low-hanging fruit for title/meta optimization.
“Show me keywords where I rank between positions 4-15 with at least 100 impressions”
Surfaces 'striking distance' keywords that could move to page 1 with focused effort.
Without GSC Data
The agent can still work without GSC connected. After crawling your site, try:
“Analyze my site structure and suggest content gaps”
Uses crawled page data to identify missing topics and thin content.
“Find internal linking opportunities across my blog posts”
The link suggester tool scans your crawled pages for relevant cross-linking.
Writing Better Prompts
The more specific you are, the better the agent's output. Compare:
"How is my SEO?"
"Compare my top 20 keywords by clicks this month vs last month. Highlight any that dropped more than 20% and suggest why."
Good prompts for the agent include:
- Specific metrics — say “by clicks” or “by impressions”
- Date ranges — “last 28 days” or “January vs December”
- Thresholds — “at least 50 impressions” or “position under 20”
- Action requests — “suggest improvements” not just “show data”
Follow-Up Questions
The agent remembers your conversation within the session (up to 20 messages). After seeing initial results, drill deeper:
- “Drill into that third keyword — what pages rank for it?”
- “Now generate a content brief for that opportunity”
- “Write an article based on that brief”
This is where the agent shines — chaining questions to go from raw data to published content in a single conversation. And when you're ready to have it run your SEO continuously rather than question by question, that's a conversation away: your free report ends with a book a call step.
Try these prompts
© 2026 Agentic SEO. All rights reserved.