As a bootstrapped SaaS founder, one of the biggest challenges I’ve faced is balancing the need for a free tier with the reality of hosting costs. If your product uses AI or LLMs, those costs can escalate quickly, making a free tier feel like a risky proposition.

Why Free Tiers Matter

These days, offering a free tier is almost a non-negotiable for SaaS startups, especially if you’re targeting early adopters. These users expect a way to test out your product before committing financially. But for bootstrappers, the risk of attracting freeloaders who rack up AI processing costs without converting to paid plans is real.

That’s the exact problem I faced with Topical, my newsletter automation platform. Topical automates the creation of newsletters by integrating with tools like MailChimp, HubSpot, and YouTube to pull in content and generate engaging email. The platform uses AI to assist with content selection and workflow setup, making hosting costs non-trivial.

Up until now, I only offered a free trial of the paid plans. But to reach a broader audience and capture early adopters, I knew I needed a more accessible entry point.

The Solution: LLM-Powered Free Tier Gate

To implement a sustainable free tier without opening the floodgates to freeloaders, I created a gatekeeping mechanism powered by a language model. Here’s how it works:

Step 1: Free Plan Request Form

When a user selects the “Free Plan” during onboarding, they’re prompted to fill out a form explaining how they plan to use the product. This is crucial because it provides valuable context that the AI can evaluate.

I show this form in the style of a "Tell us what you want to create" but you can just have a simple enquiry form if you like:

Step 2: LLM Analysis:

The submitted information is passed to a language model with the prompt:
"This user wants to create a free subscription for my product. Here is what they want to use it for. Does this seem like a valuable prospect who might one day pay for it?"

The AI assesses the likelihood that the user is a serious lead versus a casual freeloader.

Step 3: Email Domain Check:

The email address domain is also considered. Work emails are more likely to indicate genuine business use, while disposable email domains are a red flag.

Step 4: Decision Making:

Based on the AI’s assessment, the user is categorized as follows:

• High Potential Lead: Granted free tier access and invited to schedule a call to discuss their use case further.
• Moderate Potential Lead: Granted free tier access without the call invitation.
• Low Potential Lead: Denied free tier access but offered a 30-day trial of the paid plan.

Here is my full flow (click to embiggen):

Step 5: Friendly Rejection Message:

Users who don’t qualify receive a polite message like:
“Unfortunately, your use case doesn’t seem like a good fit for our free tier. However, you can still access a 30-day free trial of our Pro Plan, where our AI can help you refine your use case.”

Results So Far

Since implementing this system, about 50% of free tier requests have been granted, and they’re primarily from high-intent leads. I’ve even had some productive calls with users who were identified as high potential, leading to potential paid conversions.

By using AI to qualify free tier applicants, I’ve managed to control costs while still attracting early adopters and potential future paying customers.

For more insights on implementing LLM-powered features in SaaS products, check out this article on platform risk and how to mitigate it.

Topical is a newsletter automation platform that helps you generate, curate, and distribute email newsletters using AI. Learn more about how it works here.