Responding to Google reviews for a business

Help me Build my Workflow
1

Hello n8n community,

I’m seeking assistance in building an n8n workflow to automatically reply to 4 and 5-star Google Business Profile reviews. I’m running n8n version 1.79.3 locally via Docker, along with Postgres and Qdrant in Docker containers. I also have Ollama running bare metal and access to Gemini. I have Google Business Profile OAuth credentials but am currently waiting for my API quota to be raised from 0.

Desired Workflow Functionality:

  • Core Functionality:
    • Trigger: Schedule to check for new reviews daily.
    • Fetch Reviews: Retrieve new Google Business Profile reviews (currently using an HTTP Request node, but API quota is limiting progress).
    • Filter Reviews: Filter for reviews with a 4 or 5-star rating.
    • Generate Personalized Replies: Utilize an LLM (ideally via the AI Agent node) to create personalized replies for each filtered review.
    • Automated Reply: [Assuming a node will be added here to post replies once API access is granted]
  • Nice-to-Have Functionality:
    • Worker Name Notification: Email notification to me whenever a worker’s name is used in a generated reply. This is to allow manual review for correct spelling of worker names.
  • Separate Monthly Workflow (Optional):
    • Positive Worker Mention Tracking: A separate workflow to count positive mentions of each worker in reviews over a month for recognition purposes. Ideally, this could provide a monthly summary.

Where I Need Help:

I’m relatively new to n8n (a few weeks of experience) and am facing challenges, particularly with integrating the AI Agent node. Specifically, I’m looking for advice on:

  • AI Agent Node Configuration: Best practices for configuring the AI Agent node to generate personalized and appropriate replies to reviews. Guidance on prompt engineering within the AI Agent node for this use case would be very helpful.
  • Workflow Structure: Recommendations for the overall workflow structure, especially concerning the integration of the LLM for reply generation. I’m currently unsure how to proceed after filtering the reviews to effectively use the AI Agent.
  • Software Suggestions: Are there any other Docker-based tools or n8n integrations that could be beneficial for this workflow, particularly for enhancing the LLM’s ability to create personalized and relevant replies?

Current Progress:

I have set up a scheduled trigger and an HTTP Request node (though currently broke by API quota of 0). I also have a filter node in place to isolate 4 and 5-star reviews. I am now stuck on how to effectively utilize the AI Agent node to generate the replies.

I am open to any suggestions and happy to provide further information. Thank you in advance for your assistance!

2

Hey @nospam4chris

Try using loop along with wait option while fetching data from GMB API, I did implemented somthing similar for someone and used the same strategy

For refference you can check these details

Now when it comes to second part of your question I think the given steps should work fine

  1. Send the filtered data to an drive sheet

  2. Create an new workflow agent to write reviews (prompting is important) and generate personalised reply

  3. Create an agent to recognise worker name from the review and mail you the details (use ai agent node along with Gmail)

  4. Create another workflow agent to get worker wise data again from same sheet and consolidates the worker wise data in a separate Google sheet (good prompt is important) :grinning:

  5. Connect all the sub-workflow with a main agent

I hope this would help you resolve some problems

If you need any further help do let me know

4

Hi Chris, could you get this agent running? I just found your one year old post while researching for the exact same thing. I would appreciate any feedback which could be helpful. thank you Markus