Greg Mulholland, the founder and CEO of the machine learning materials startup Citrine Informatics, visited Cyclotron Road last month as part of our founders’ speaker series. These talks feature current and former technical founders discussing their experiences starting hard technology companies. Although technically a software startup, Citrine has had a huge impact in the materials manufacturing industry, where its platform has reduced the time to market for new products by 50%. It is thus an amazing example of how to capitalize on opportunities situated at the intersection of hard science and software innovations.
We spent two hours with Mulholland in an engaging and informative discussion that touched on everything from defining value proposition to making strategic pivots to the importance of a support network for stressed startup founders. Here are three takeaways from our conversation.
1. Product presentation can make all the difference
The biggest strategic change that Citrine made, according to Greg, was not a change in product but a change in presentation:
“When we first got started, we would just say ‘hey give us your data, we’re gonna be smart with it and… it’s gonna make you better at science!’ Turns out… they [clients] don’t have the data to give and it’s a bit of a hard sell because who are we to say we’re better at polymers than [the major manufacturers in the field]? We’ve had to really refine how we talk about business in a big way… the business has pivoted a little bit, but I think the way we present it has changed a whole lot.”
Citrine’s clients include large and well-established manufacturers. One of the things that Mulholland realized early on was that these companies didn’t have that much data - and the data that they did have included old paper notes sitting in warehouses. Although Citrine had initially positioned itself as a platform to improve the materials discovery process using machine-learning, it was forced to reorient itself to a new position, which included data extraction and management.
“We had to realize very early on that companies don’t have that much data - and the data they do have is [in an awful format.] How do you get these [paper] technical reports into a system that lets us do machine learning? Our door to these companies became the ability to extract that data, so now when we go into a company, we go in saying ‘your data is trapped in documents and we can take it out and make it useful.’”
2. There is a huge market in doing the things that people hate
Mulholland began by talking about Citrine’s first hypothesis; that it needed to be the center of attention for all things data, in order to differentiate itself from free software competitors like python. However, the way in which this translated into value for customers was not immediately clear, and came about from doing what Mulholland called ‘lean, startup-style things.’
“The first lean startup style thing we did was go talk to [a scientist at] Applied Materials - and the thing she wanted more than anything was an on-demand literature review of batteries immediately, any time she wanted…. Really, what she wanted to know is what does the frontier of my field look like today? And because we have this singular focus on data, we actually will go get those literature reviews.”
This led to an important business learning; namely that performing services that address a major pain point can serve as a great way to first hook a client and then position your product.
“There’re 3 tiers of things a customer will do - there’s the things that they absolutely want to do themselves… things they want to own and they’re the best at … and then there are the things that they can do but are a complete pain - collecting data, structuring and sharing data… and the third is things that they will never do in a million years. So what we’ve said is ‘Look we’re not the ceramic beads people, we’re not the polymer people, what we are is the tool that helps all of those people do their job better, because we really love doing the things you hate to do...
The problem that a materials developer has is not the materials - materials are tools, an endpoint. The problem is the pathway from where I am today to what materials I need. And so for us, the value is in time to market reduction.”
As an example, Mulholland noted that Citrine’s first deal with an early client was for data extraction, but the client now uses Citrine for all things data and machine learning related. Even though their core technology remained unchanged, the way in which they presented themselves, and the tasks they were willing to perform changed the way they positioned themselves along with their value proposition
3. Startups are hard, and having a support system is important
Mulholland concluded our discussion with a personal perspective on what it was like being a startup cofounder, reiterating again and again that working on a startup is a difficult, exhausting process. When faced with multiple deadlines, long hours, and exponentially increasing caffeine consumption without an outlet to decompress, the toll on your emotional health can be devastating. Having the right support network, especially when it consists of other founders who are able to empathize with the struggles of running a startup, can alleviate this stress significantly.
“Everybody should have a hotline, like your own therapist… someone you can talk to in a judgement-free zone. When we were at StartX at Stanford, we had a lot of other teams around us we could share stuff with, and now we kinda just don’t have that - we’re at our own office, we’re kind of our own standing entity, and we don’t see each other as much…
For me, finding those few moments of support were huge in terms of confidence and feeling like I was really building something. I can look back now at where I was three months ago and say wow that was really terrible, but you want to make sure that when somebody is in those depths that they have somewhere that they can go”