9 ways to burn money before you launch your product

Rachel Draelos is an MD and PhD. In 2018, in the middle of her training, she started building Cydoc, an AI startup to help doctors document visits. Seven years later, in August 2025, she shut the company down. On February 21, 2026, she published a detailed postmortem on the Glass Box Medicine blog, with numbers and nine mistakes.

I read the whole thing and recognized many conversations I've had with founders. Her numbers are small, and that's why they're useful: they show how you can lose years without ever spending a fortune at once. Below are her nine mistakes, adapted for anyone about to ship a product.

Cydoc's numbers

• 7 years as solo founder and CEO.
• US$265k raised from family, friends and doctors, after pitching more than 35 funds.
• First revenue in December 2023: US$190.36.
• Total revenue in 2024: US$27.9k.
• 4 paying customers at peak.
• The product saved about 10 minutes per visit, according to real customers.

The product worked. The business didn't add up. That gap is what this piece is about.

1. Talking to customers too late and too little

She thought she already knew what doctors wanted. She talked to few real customers, and late. Today she recommends at least 20 interviews, ideally more than 100. My favorite line of hers: “Nothing shows real customer interest like cold hard cash.”

2. Building an MVP that's too big

Cydoc launched complete, polished and late, with interfaces no customer used. It's the mistake I see most. The founder calls something an MVP that would take a year to build. An MVP is the smallest product someone would pay to use, not the first version of the dream.

3. Developing without a process

Early on, she rewrote code for every interface variation and had no standard engineering practices. Every change cost double. Without a process, speed at the start turns into slowness in the middle.

4. Waiting for customers to show up

She assumed customers would come and bet that adding an AI scribe would make selling easier. It didn't. A new feature doesn't replace someone selling.

5. Doing everything alone

As a solo founder, she held the roles of CEO, CTO and head of revenue. Sales and marketing got no attention. If you don't have a technical cofounder, the problem isn't just building. It's having time to sell while someone builds.

6. Avoiding the integration customers needed

Instead of integrating with the electronic health records doctors already used, she tried to build her own. Each integration would have cost between US$4k and US$6k, but without them adoption was too hard. Customers won't swap the system they use every day just to try yours.

7. Trusting a moat that didn't protect

She counted on two patents and the complexity of the code. A customer built a simpler version of the core idea. In her words, Cydoc's moat “was more of a rain puddle.”

8. Falling into the custom-work trap

A psychology clinic dangled a US$5k-a-month subscription. She did custom work, for free, on a flow with more than 200 questions. The subscription never came. A promised contract is not a contract.

9. Seeing too late that the model didn't work

Hosting and AI cost about US$70 per doctor. Small clinics paid less than US$100 a month. That left roughly US$30 in margin, before any integration. Her math: 11 years to break even on each clinic. By the time it was clear, there was almost no money left.

The pattern behind all nine

Looking at the list, almost every mistake shares one root: time and money spent building before there was proof anyone would pay. A big MVP, the wrong integration, free work and a model that doesn't add up are all versions of the same postponed decision.

That's why I push so hard on dates. A fixed date forces you to choose what's in and what's out, and it forces you to put something in the customer's hands early. Without a date, every mistake on the list has time to grow.

What this means for you

Before you invest in another month of development, answer:

1. How many customers have paid or committed to pay?
2. What in your product did no customer ask for? Cut it.
3. Are you doing custom work without a contract?
4. What does it cost to serve a customer, and what do they pay? Does the difference add up?
5. What's the date the product will be live, and what has to come out of the scope to make it real?

How we do it at the Studio

The price is fixed after discovery, and the deadline is written into the contract. In discovery, the main question isn't “what can we build” but “what's the smallest product that proves the business”. AI Vou Eu, an AI travel SaaS, is one example: with the scope cut to fit the date, it went live in 4 weeks.

If you want to launch a system or SaaS without repeating these nine mistakes, start with discovery in our custom system and SaaS development service.