Former Stanford professor Peter Bailis raised $66M for his data startup, Sisu. This article breaks down the tactical lessons from his journey: how to commit fully to your idea, fundraise for a complex product by de-risking the technology, and strategically select your first customers to ensure long-term success.
Key takeaways
- Go all-in; part-time commitment signals risk to investors and early hires.
- De-risk a technical pitch by validating the problem with industry partners *before* you start.
- Your first customers aren't just for revenue; they are design partners who shape the product.
- Tell a story about market pain, not just technology, to win over top VCs.
- Use a pilot program to prove value and secure lighthouse customers early on.
Your Data Is Telling You What's Wrong. You're Just Not Listening.
You have dashboards. You have metrics. You have more data than ever. Yet when a key metric like user retention, conversion rate, or churn suddenly changes, you're blindsided. It kicks off a frantic, all-hands scramble to figure out why . This is the dirty secret of the modern data stack: most companies are great at seeing what is happening, but clueless as to why it's happening.
Peter Bailis, who became a Stanford computer science professor at just 25, saw this problem firsthand inside the world's top tech companies. He left a coveted academic career to found Sisu, a company that helps businesses understand not just what their metrics are, but why they're changing. After raising $66.7 million from top firms like Andreessen Horowitz and NEA, his journey offers a tactical playbook for technical founders.
This isn't just a story. It's a guide to making the leap, raising capital for a complex product, and finding the first customers who will make or break your business.
The Leap: Are You Tinkering or Building a Business?
At Stanford, faculty members can take a two-year leave to explore other ventures. Many great companies have started this way. It's a tempting, seemingly safe path for an academic or expert with a good idea. You can keep one foot in the safety of a prestigious job while dipping a toe into the startup world.
This is a founder's most common, and most dangerous, first mistake.
Bailis faced this exact choice. He could have used his allotted one day a week or taken a formal leave to "tinker" with Sisu on the side. Instead, he forced himself to answer a more critical question:
When I'm on my deathbed, am I going to regret having taken the safe path? Or will I be proud that I jumped in with both feet and built something from scratch?
He resigned and went all-in. For you, the founder, this decision point is critical. Part-time commitment is a signal of risk to everyone: to the co-founder you need to convince, to the first engineer you need to hire, and especially to the investors you need to fund your vision. If you're not fully convinced, why should they be?
The takeaway: The "safe" path of tinkering on the side feels rational, but it severely limits your upside and telegraphs a lack of commitment. The startup clock doesn't truly start until you are 100% focused.
Fundraising for Deep Tech: The $66M Story
Sisu raised a staggering $66.7M from top-tier investors. The source material credits this to a massive market, enterprise demand, and a strong team. This is true, but it's table stakes. It misses the real lesson in how they de-risked the company for investors before the first pitch.
For a technical founder, especially one coming from academia, the biggest risk is not the technology; it's the market. Can you build it? Yes. Does anyone care? Maybe.
Bailis didn't start by writing code for Sisu. He started by building credibility and validating the problem through his academic work at Stanford. The DAWN Project, his research group, wasn't just theoretical. They secured funding and partnerships with industry giants like Google to work on real-world problems, improving the quality of their ads and the usability of their data warehouses. He found that even at Google and Facebook, machine learning was in its infancy in terms of accessibility.
How to De-Risk Your Technical Vision
Find Industry Partners, Not Just Advisors: Bailis's team didn't just get quotes for a website. They got companies to fund open-source development. This is the strongest form of pre-commercial validation. · Solve a Real-World Problem: Their research focused on making data science more usable, not just more accurate. This focus on application, not just theory, demonstrated a customer-centric mindset that is rare in academia. · Build a Pre-Seed Track Record: By the time he approached VCs, Bailis wasn't just a professor with an idea. He was the leader of a well-known research lab that had already delivered value to FANG companies. His open-source projects (Weld and Snorkel) were, in effect, his first MVPs.
The storytelling that closed Ben Horowitz wasn't just a 20-slide deck. It was the culmination of years of work that proved the problem was massive and that Bailis was one of the few people in the world who could assemble a team to solve it.
How to Select Your First Customers
The original article mentions this as a key topic but provides no details. For an enterprise company like Sisu, this is the most critical early-game decision. Your first five customers are not just revenue; they are your unofficial product team, your marketing department, and your reference-able proof points for your Series A.
Before starting Sisu, Bailis's team was already "giving away a taste of it and getting others to try it." This is how you should approach your go-to-market. Frame it as a highly selective "Design Partner Program." You are not desperately looking for anyone to use your product; you are curating a small group of experts to help you build the future.
Checklist: Your Ideal First 5 Design Partners
Don't just sell to whoever answers your cold email. Be rigorous. Your ideal design partner fits this profile:
They Have a Hair-on-Fire Problem: They are actively and painfully aware of the problem you solve. They've tried to build a solution internally, and it failed. You don't have to convince them the problem is real; you only have to convince them your solution is better. · They Have the Right Resources: For Sisu, this meant customers with massive, clean datasets and the engineering talent to support an integration. For you, it might mean budget, a dedicated point of contact, or executive sponsorship. A partner who can't implement your solution is useless. · They Can Become a Lighthouse: Is their logo one that will make other customers in your target market take notice? A no-name customer paying $1,000/month is a distraction. A well-known brand in a key vertical paying nothing for a 3-month pilot is a strategic asset. · You Have a Champion: You need an internal evangelist who will fight for your product, navigate procurement, and get you access to the right people. Your sale is not to a company; it's to a person. Find them and make them a hero.
Common Mistake: Charging Too Little (or Too Much)
Founders agonize over pricing for their pilot program. Don't. The price is not about revenue; it's about commitment. Charging nothing signals your product has no value. Charging a nominal fee (e.g., $1k-$5k/month for a product that will eventually cost $100k+) is the right approach. It forces the customer to get budget approval, which validates their commitment and ensures you're working with a real stakeholder.
How to Apply This Next Week
Run the "Deathbed" Test: Are you all-in? If not, what specific milestone will it take for you to quit your day job? Write it down. Be honest with yourself. · Map Your Pre-Founder Validation: Make a list of every project, open-source contribution, or consulting gig you've done that proves this problem is real. This is the foundation of your pitch's "Why Now" and "Why You" slides. · Define Your Ideal Design Partner: Use the checklist above to write a one-paragraph description of the perfect first customer. Get specific about their industry, size, and the title of your champion. · Draft a Design Partner Outreach Email: Don't ask to "pick their brain." Offer immediate, specific value.
My name is [Your Name], and I'm the founder of [Your Company]. I'm reaching out because I saw [Their recent launch, a job posting for a data analyst, etc.] and I've spent the last [X years] focused exclusively on helping companies like yours understand the 'why' behind their metrics.
We're building a new platform to automate this process and are selecting a small group of 5 design partners to help us shape the product. Given your role, I thought you might have insights.
Would you be open to a 15-minute call next week to see if it's a fit?
Frequently asked questions
- How much should a deep tech or AI startup raise?
- It depends on talent costs and sales cycles. A large seed ($3-5M) or Series A ($10-20M) is common for capital-intensive R&D. Sisu's $66M was raised over multiple rounds, including Series A and B.
- How do I transition from an academic or technical role to being a founder?
- Start by validating the commercial problem, not just the technical solution. Talk to at least 50 potential customers. Use Peter Bailis's 'deathbed' test to decide when to commit 100%.
- What do VCs look for in a technical founder?
- They look for your ability to connect your technology to a massive business problem. They want to see that you can recruit a team, sell a vision, and are obsessed with solving customer pain, not just building cool tech.
- What is a 'design partner' and how is it different from a customer?
- A design partner is an early customer who agrees to provide deep feedback to help you build the product, often in exchange for preferential pricing or terms. They are more collaborative than a typical transactional customer.
