AI Consultant

The AI Consultant is a chat that behaves like a marketing consultant: it looks at your data, proposes a plan, and — once you approve — executes the work inside Tuurbo.

What makes it different from a generic chatbot

• It reads your audit results, Search Console data and Knowledge base before answering.
• It produces action plans, not just advice: ordered steps with expected impact and credit cost.
• It can execute the approved steps (generate meta tags, rewrite products, run audits, enable automations).

Step-by-step: from question to result

1. Open a thread. Every thread keeps its own context, so use one per topic ("winter collection", "site speed", "German market").
2. Describe the goal, not the task. "We want more organic traffic on running shoes" works better than "write 10 meta descriptions".
3. Let it analyse. It pulls the latest audit and metrics for the current website.
4. Read the action plan. Each step shows what will change, where, and how many credits it costs.
5. Approve, edit or reject steps. You can approve one step at a time.
6. Watch execution. The Consultant reports progress and links to the changes it produced.
7. Follow up in the same thread. Ask "what changed?" or "what is next?" and it continues from where it stopped.

Skills

The Consultant has skills for specific jobs: keyword research, competitor comparison, content briefs, product content, technical SEO diagnosis, accessibility review, reporting. You can invoke one explicitly, or just describe the problem and let it pick.

Good prompts to start with

• "Audit our top 20 landing pages and tell me the three biggest problems."
• "Build a 30-day plan to improve visibility in AI answer engines."
• "Which product pages lose the most impressions to weak titles?"
• "Prepare a monthly report for a non-technical stakeholder."

Credits

Each message costs 1 credit. Operations launched from a plan cost their own price (for example 5 credits per product description). The plan always shows the total before you approve.

Limits to keep in mind

• It works on one website at a time — switch project to change context.
• It can only write where a connector or the script gives it permission.
• An empty Knowledge base produces generic copy; fill it first.