Using the Leat AI Assistant with Claude and ChatGPT
Leat's AI Assistant connects your loyalty program directly to Claude, ChatGPT, Perplexity, and other AI tools, so you can ask questions about your program instead of clicking through dashboards. This lesson covers how to connect it and put it to use.
Key Takeaways
The Connector URL: Installing the AI Assistant from Apps > Integrations in Leat generates a connector URL. This single URL is what links your account to Claude, ChatGPT, Perplexity, or any other AI tool, and is pasted in as a custom connector on the AI tool's end.
Authorization via Login: Adding the connector in Claude prompts you to log in to Leat and click Allow. This step is what grants the AI tool permission to read your program's live data.
Direct-Question Analysis: Once connected, prompts like "Please analyze my loyalty program performance from Leat" replace manual dashboard navigation, with follow-up questions narrowing into specifics like a month-on-month transaction breakdown or a chart of that data.
Balanced Suggestions: Asking AI more specific questions like what reward to configure next prompts it to look at where customer point balances actually cluster, rather than just the average, and flag gaps in your existing reward tiers.
Dashboard Checklist
In Leat, go to Apps > Integrations. Select Leat's AI Assistant > Install, and copy its connector URL at the bottom of the page.
In Claude, go to Settings > Connectors > Add custom connector, name it, and paste in the URL.
Not using Claude? For ChatGPT, navigate to 'Plugins' > 'MCPs' > 'Add server' and enter a name and select 'Streamable HTTP'. Paste your connector URL from Leat, click 'Save,' and return to the MCP server list and click 'Authenticate.'
Log in to Leat when prompted and click Allow to authorize the connection.
Start a new chat and ask AI to assess any aspect of your loyalty program.
Prompts to Get Started with Leat's AI Assistance
Connecting Leat to an AI assistant like Claude, ChatGPT, or Gemini means you can skip the dashboard-digging and just ask. Below are some examples of prompts you can copy, paste, and adapt to your own program, split into two categories: understanding how you're doing today, and figuring out what to do next.
Swap in your own reward names, timeframes, and channels wherever you see brackets.
Understand Your Performance
Get the full picture
Give me an overview of my loyalty program's performance over the last [30/60/90] days. Include enrollment, points issued vs. redeemed, and repeat visit rate.
Spot your winners
Which rewards have been redeemed the most in the last [month/quarter]? Are there any rewards not being actively redeemed that I should consider replacing?
Find your best customers
Who are my top 10 most loyal customers by visit frequency or spend? What do they have in common?
Check the health of the funnel
What percentage of customers who earn points actually come back for a second visit? How does that change for a third or fourth visit?
Compare time periods
How does this month's loyalty activity compare to last month? Flag any significant changes.
Plan What's Next
Get reward ideas
Based on my current rewards and how customers are redeeming them, suggest 3 new reward ideas that might increase repeat visits.
Test a channel expansion
I currently run loyalty through [in-store / online / both]. Based on my customer activity, would expanding to [new channel] likely help retention? What would I need to set up?
Design a win-back campaign
Identify customers who haven't visited in [30/60/90] days but were previously active. Suggest an offer that might bring them back.
Plan a seasonal push
Suggest a loyalty campaign for [upcoming holiday/season] using my existing reward structure. Keep it simple enough to launch in a week.
Sanity-check a new idea
I'm considering [describe your idea, e.g. a tiered rewards system, a referral bonus]. Based on my current customer data, does this seem like a good fit? What would you want to see to be more confident?
Tips for Better Answers
Be specific about timeframes - "last quarter" gives a more useful answer than "recently."
Ask follow-up questions - treat it like a conversation, not a single query. "Why do you think that reward underperforms?" often surfaces useful detail.
Combine performance and strategy - e.g., "Given [X result], what should I try next?" gets you from data to action in one step.
Once your Leat account is connected, any of these prompts can be pasted directly into your assistant to get started.










































