How we signed a world-famous AI company with a HackerNews post

Landing your first big customer is a rite of passage for startups. If you’re an early-stage company, bigger customers want to see that you’ve worked with others like them. This is even more true if you sell infrastructure your customers trust for important parts of their business like billing.

‍

But getting those customers means you’ll face the dilemma of a recent graduate who can’t get a job for their lack of experience and can’t get experience for their lack of a job: You can’t get the big reference customers because you don’t have big reference customers.

‍

And even when you get a demo scheduled, enterprise procurement is arduous: It often takes months (or even years) from the first demo to closing. With our first big customer Mistral AI, it took 17 days. Less than 2 months after they first inquired, we were live in production. 

‍

Getting such a high-profile customer early has accelerated our business and helped us raise our Series A funding

‍

Here’s how it went. 

How we signed Mistral AI from a Hacker News post

‍

In February of 2023, we posted about Lago on HackerNews:

The post itself got a lot of engagement. But we didn’t know yet just how well it would work.

‍

Because 8 months after posting, we received a form submission on our website from a world-famous company: Mistral AI. 

‍

Mistral had just raised their $113m seed round a few months earlier and was starting to productize their models and needed billing. Through a form on our website, Mistral’s head of engineering told us: 

‍

What’s your pricing model? Token-based 

How do you bill customers? We don't bill our customers yet 

What are the issues with your current billing system? We don’t have one yet :-)

‍

We started chatting and moved quickly. The biggest reason for that was that we trusted each other right away, which matters a ton for anything infrastructure-related. And I believe that as an early-stage company, we built trust with Mistral for a few key reasons:

• First, we both started in Paris. Remote collaboration is great, but there’s something to getting together in person that gives you a better feel for who you’re dealing with and builds trust faster. This is also why I moved to San Francisco to start expanding our presence in the U.S: We saw that about 30% of both users and customers were located in the U.S. (before we had any presence in the U.S.), which was a strong signal that we should be there.
  ‍
• Second, working with us was low-risk because we don’t believe in lock-in. As an open-source platform, Mistral could always choose to self-host, inspect our code, and gain full visibility into how we built our product. This transparency not only built trust but also enabled Mistral to understand our code deeply, making it easier for them to build on top of Lago. If they ever needed more control over their infrastructure, self-hosting would always be an option. Additionally, we integrate with multiple payment vendors, providing them with greater flexibility moving forward compared to a more closed system.
  ‍
• Third, we were always willing to go the extra mile. Stripe went after Mistral at the same time we spoke with them. But Stripe Billing only works with Stripe Payments, which would’ve meant being locked into Stripe. Lago integrates with Stripe Payments*, which gave Mistral more flexibility if they needed to change vendors in the future. 

‍

My main takeaway here is that, as an early-stage startup, few things matter more than the relationships you have with your customers. If you’re buying a product from an early-stage company, you need to be sure that a) the company will be around for a long time and b) the product will grow into the perfect product for you. 

‍

Speaking of the product, we also committed to some product decisions for Mistral. 

Why your customers should contribute to your roadmap 

If you’re a frontier AI company like Mistral, you need to serve customers worldwide. As we started talking with Mistral, they were already looking to start entities beyond their original French one so they could serve customers anywhere. 

‍

One of the biggest issues with this is taxes. In the EU alone, there’s a lot of complexity around VAT. Once you go outside of the EU, there’s additional complexity. 

‍

While we didn’t have these tax features in the beginning, we committed to building them quickly for Mistral. Whether you should build features for a specific customer is always a difficult decision. 

‍

On the one hand, you want to pursue your own product vision instead of being a feature factory. If you only build what prospective customers ask you for, you’ll just replicate the product of the incumbents you’re trying to disrupt. And as an early-stage company, your opportunity cost is greater because you don’t have hundreds of engineers at your disposal.

‍

But if you’re a founder, you’ve probably built things you were excited about, but your users weren’t interested in at all. By building what customers ask you for, you build something people want by definition. Since then we’ve made our roadmap public and participative (see here).

‍

My thinking around it goes something like this: If a customer/prospect requests something that a) we would’ve built anyway and b) will benefit other customers too, then we should build it. 

‍

From our time building billing at Qonto, we knew we would eventually build the tax tooling, so it made sense to pull it forward and put it on the roadmap right away. 

‍

The above is more about product strategy, but it goes back to the aspect about trust. 

How to sign your first big customer

To recap, I think there are a few big lessons to be learned here. 

‍

1. First, don’t get too fancy about your marketing right away. It would’ve been easy to spend weeks and thousands of dollars on a fancy video, but we “just” wrote about our story and product on Hacker News. Early on, the opportunity cost of delaying is much bigger than the upside of perfecting your marketing materials.

‍

2. Second, once you’re talking to a (prospective) customer, you need to build trust with them. The more you have in common, the better. Mistral and us are both companies with big, international ambition and French roots—and both believe in open source. 

‍

3. Third, be willing to pull forward roadmap items and open to adding new ones, especially if it also benefits other customers. 

‍

‍

Read more here:

• ‍How open can Stripe be?
• ‍We crunched the fees of Stripe’s 21 products
• How Lago uses Lago: Tackling our owns Billing nightmares

‍