AI Research Interviewer Agent using n8n, Groq, and Redis

Unlock In-depth User Insights with this AI Interviewer Agent

This AI Agent automates the user interview process, acting as a sophisticated researcher. It initiates an interview via a simple n8n form, then dynamically generates relevant, open-ended questions using a Groq-powered LLM (e.g., Llama 3.2 90b) orchestrated by a Langchain agent. The agent engages users in a continuous dialogue, adapting to their responses and digging deeper when appropriate. All interactions are stored in Redis for session management and later transcribed to Google Sheets for comprehensive analysis, empowering you to gather valuable qualitative data at scale and on demand.

Key Features & Benefits

• Automated User Interviews: Conduct interviews 24/7 without manual intervention, saving significant time and resources.
• Dynamic Question Generation: Leverages Groq's LLMs via a Langchain Agent to ask contextually relevant and follow-up questions, ensuring rich data collection.
• Session Management with Redis: Maintains interview context using Langchain memory buffers and stores full transcripts in real-time for each user session in Redis.
• Transcript Recording & Export: Automatically saves full interview transcripts to Google Sheets for easy access, sharing, and analysis by your team.
• Customizable Interview Topic: Easily configure the core topic for the AI to focus on during the interview.
• User-Controlled Interaction: Allows users to end the interview at any point by typing 'stop interview' or a similar phrase recognised by the LLM.
• Optional Transcript Display Page: Includes a webhook-powered HTML page to display the full interview transcript to the user upon completion.

• An n8n instance (Cloud or self-hosted).
• Groq API Key with access to a suitable model (e.g., llama-3.2-90b-text-preview).
• Redis instance credentials (e.g., from Upstash.com, Aiven, or self-hosted).
• Google Sheets API credentials configured in n8n for saving transcripts.

1. Download the n8n workflow JSON file.
2. Import the workflow into your n8n instance.
3. Configure the 'Start Interview' (Form Trigger) node: customize the formTitle, formDescription, buttonLabel, and path as needed for your interview.
4. Configure all Redis nodes ('Create Session', 'Update Session', 'Update Session1', 'Update Session2', 'Get Session', 'Query By Session') with your Redis credentials.
5. In the 'Set Interview Topic' node, update the interview_topic variable with your specific research subject. This topic is fed to the AI Agent.
6. Configure the 'Groq Chat Model' node: enter your Groq API Key and ensure the selected model (e.g., llama-3.2-90b-text-preview) is appropriate for your needs.
7. Review the systemMessage in the 'AI Researcher' (Langchain Agent) node. Adjust its instructions if necessary, but ensure the JSON output schema (stop_interview, question) is maintained as the workflow depends on it.
8. Configure the 'Save to Google Sheet' node: select your Google Sheets credentials, provide the Spreadsheet ID and the Sheet Name where transcripts should be appended. Review the column mapping if needed.
9. (Optional) Configure the transcript display feature: a. The workflow includes a 'Webhook' node (default path: ai-interview-transcripts/:session_id) to display the transcript. After activating the workflow, copy the Production URL of this 'Webhook' node from your n8n instance. b. Go to the 'Redirect to Completion Screen' (Form) node. c. In its redirectUrl parameter, you need to construct the URL that points to the 'Webhook' node dynamically. Use the Production URL you copied in step 9a, and ensure the session ID is correctly appended. The value should look like: YOUR_COPIED_PRODUCTION_WEBHOOK_URL_BASE/{{ $('UUID').first().json.data }}. For example, if the production URL for the path ai-interview-transcripts/:session_id is https://your.n8n.instance/webhook/transcript-viewer/ai-interview-transcripts/test, the redirectUrl should be set to https://your.n8n.instance/webhook/transcript-viewer/ai-interview-transcripts/{{ $('UUID').first().json.data }}. Adjust the path based on your n8n webhook URL structure.
10. Activate the workflow. Test by accessing the Form Trigger URL provided by the 'Start Interview' node.