Steps to build your own customer knowledge center
This resource is based on How to build an AI brain from your customer data in less than 2 hours, featuring Sean Blanda of Gate Check Studios, published on the AI Lab by ActiveCampaign.
Get the checklist
Before you start
- ChatGPT Plus or Pro account (Projects requires a paid tier)—or Claude Pro
- Access to your customer-facing tools (review sites, support software, email, call recordings)
- Export permissions for the tools you plan to pull data from
The workflow
Phase 1: Track down your customer inputs (~30–60 minutes)
- List all digital touchpoints: where do customers talk to or about your business?
- Gather customer reviews: Google Local, Yelp, G2, Angie, industry-specific review sites
- Pull support conversations: Zendesk, Intercom, Help Scout, or shared support inboxes
- Collect email communications: customer replies, support threads, feedback emails
- Export social media interactions: Buffer, Hootsuite, or platform-native exports
- Gather call transcripts: Zoom, Google Meet, Granola, Fathom, or Gong recordings
- Pull survey responses: post-purchase surveys, NPS forms, feedback forms (Typeform, SurveyMonkey)
- Check your marketing platform: ActiveCampaign or similar tools for email response data
Phase 2: Capture and export the data (~30–60 minutes)
- Export Google reviews: use Google Takeout for local business reviews
- Export call transcripts: download transcripts from your call recorder as text files
- Export support data: use your support tool’s export function (text or CSV)
- Export survey results: Typeform, SurveyMonkey, or your survey tool’s export feature
- Export review site data: Trustpilot, Yelp Business, or manual copy if no export exists
- Save everything as text or CSV files: raw data is fine; AI handles the structuring
Phase 3: Create your AI project (~10–30 minutes)
- Create a new project: ChatGPT sidebar → “New Project” (or Claude → “New Project”)
- Name the project: e.g., “Customer Knowledge Center” or “[Business Name] Customer Brain”
- Upload all exported files: select “Add files” and upload your text/CSV exports
- Test with a question: ask something specific like “What are common words my customers use when describing [specific process]?”
- Add custom instructions: click “Add instructions” and include your style guide or brand voice rules; competitors to never mention; “cite all quotations and customer feedback with the source document and name of the customer”
- Share with your team: use the project’s team functionality so coworkers can query and upload
Quick reference
- Total time: 1–2 hours
- Tools needed: ChatGPT Plus/Pro (or Claude Pro), export access to customer-facing tools
- Key output: A queryable AI knowledge center that surfaces customer language, testimonials, patterns, and insights from all your feedback data
Ready to read the full story?
Read How to build an AI brain from your customer data in less than 2 hours, featuring Sean Blanda of Gate Check Studios, published on the AI Lab by ActiveCampaign.
Related
More data from the AI Lab.