Ever feel like software pricing is stuck in the past? I sure do. The flat monthly fee model has dominated SaaS for years now. It's predictable for both sides but creates some weird misalignments. You charge the same whether a customer uses your product once a month or a thousand times daily. Doesn't quite feel right, does it?
But there's a revolution brewing, and it's coming from an unexpected place: artificial intelligence.
My Token Education
I recently started experimenting with Lindy.ai to build some automations. The experience has been eye-opening, not just for what the AI can do, but for how they charge for it.
Here's what Lindy.ai handles for me now:
- An email comes into my inbox and the AI extracts the details and appends a tracking sheet
- I need to schedule a meeting? I just copy my AI agent into the email request
- Need research on someone I want to speak with? The AI searches, scrapes data, and compiles it into a brief
Am I pushing Lindy.ai? Not yet. There are functions it doesn't do well. As with all AI offerings, you need to learn what it's really good at and then assess if it provides enough value to stick with it.
But what I want to talk about isn't the AI capabilities themselves—it's the pricing model, because that model may be the opportunity for all software vendors to rethink how they charge.
Let's Talk Tokens
Yep, that same phrase that might have dulled your eyes when speaking about crypto IS the model for AI. But in this context, it's actually fascinating.
For AI, a token is a small amount of characters (think 12-18) that add up to the text, numbers, images, sounds, etc. that form the input and output of the AI. In other words, AI doesn't ingest big things as wholes—it breaks them into many small chunks of data and then analyzes all the pieces separately and together. It then builds those small things back together into what you want—ad copy, financial graphs, images, videos, etc.
Each token has a compute cost, and those costs create a price for tokens.
The Pricing Breakthrough
In the case of typical consumer GPT tools where you enter commands into a screen and get results back, AI companies often charge a flat monthly fee. It seems $20 a month is where things are settling. But here's the open secret: if you're even a light user of those systems, it's costing the AI providers significantly more money to serve you than you're paying.
When you don't use the command prompt and instead pay for compute via APIs, you pay by the token. And that's where things get interesting.
When you charge by the token, you as the software company are 100% aligned with the customer in providing value. You want to be used for any possible use case that delivers value. As the platform, every token provides a minute profit. To use an old web 1.0 term, it's microtransactions, and those transactions build up fast.
Want 1080p video instead of 480p? You double token consumption. Want an article? That's 100 tokens. Want a book? That's 1.2 million tokens.
Token Pricing in Practice
Lindy is charging 1 penny per token for casual users with a 5,000 token minimum. I can use default automations or create custom ones. When I parse a receipt, 3 tokens are consumed, which is 3 cents. When I have Lindy negotiate a meeting time and set appointments, it's generally 270 tokens consumed, or $2.70.
This is just 2 of literally hundreds of thousands of automations I can create. Upgrading is easy. I just buy more tokens.
So Why Are So Many SaaS Companies Stuck in a Flat Fee Model?
It's because it provides predictable revenue. It also creates tremendous sales expense to grow rapidly with a single product. The alternative—adding multiple products—is once again an entirely new sales effort.
But token-based pricing changes all that.
Applying Token Economics to Traditional SaaS
Imagine you sell a point-of-sale system. Instead of charging $110 a month, you sell 10,000 tokens a month:
- Process an order? That's 25 tokens
- Clock in/out? That's 1 token
- Display on a kitchen display system? That's 5 tokens
Token use will scale up or down with use. If the business has seasonality, it'll consume variable tokens. If you want to introduce back office, labor and scheduling, or other SaaS functions... they just consume tokens.
There are no new contracts. There is no sales effort other than to encourage more valuable usage. If customers don't find value in a particular feature, they simply stop using that part of the software.
The Enterprise Advantage
This is where it can get really interesting for larger clients. Instead of discounting per location or per user, you simply give them more tokens for each dollar they give you. You can offer to bank some or all unused tokens in a given month or year. You can also pool tokens across locations.
If you sell to a large company like RBI, which owns Burger King, Popeyes, and Tim Hortons, you can offer a system pool of tokens. If you want to add resellers, you just have them buy tokens at a lower price, and they upcharge with a push for them to buy a pool which they can allocate to their clients.
The revenue model becomes flexible and hard to replicate competitively. It's easy to find out how much per location someone pays for SaaS, but it will be difficult to know how many tokens are consumed to generate a custom report.
The Future is Token-Based
AI is taking over software. It seems natural that the next generation of startups will be AI-first SaaS companies. I see no reason why they won't deploy a token-based revenue model.
Maybe your existing SaaS company should consider the same. After all, wouldn't you rather have customers who pay for exactly what they use and value, rather than those who pay a flat fee regardless of whether they get real benefit?
Token economics isn't just for AI companies—it's the future of software pricing. And the companies that adapt first will have a significant advantage.
Are you ready to tokenize your business model?