• Mike Leslie

Fundraising for an MLOps Company

Updated: Jul 4

We wanted to share some of our experiences while fundraising for our Machine Learning Operations (MLOps) start-up Revela. There are a few factors that make fundraising for this type of business unique.


First you might ask, what exactly is MLOps? As defined by DataBricks, MLOps is a “core function of Machine Learning engineering, focused on streamlining the process of taking machine learning models to production, and then maintaining and monitoring them.” Just like DevOps is to software development, MLOps is all about the processes and systems required to get Machine Learning models out of development and into everyday products. We've discovered that pitching an MLOps company to investors has some challenges, but also some advantages.


Connecting Investors to our Customer Problem

One of the biggest challenges around fundraising for an MLOps company is that we are almost certainly one or more levels removed from an investor’s direct experiences. Unless an investor has a background in Machine Learning, there will be no “A-hah!” moments where they can relate to the problem we are solving for our customers. On the other hand, if you're a start-up launching an App that streamlines the booking process for salons, almost every investor you talk to has had a haircut and has lived through the process of booking a haircut. They may or may not agree on the viability of the business, but they can at least directly relate to the customer problem.


There are usually at least two layers of separation between the problem we are solving and the investor’s experiences. For example: Almost everyone has used software like Google maps for getting directions, so they can relate to the value of having very accurate time estimates for arriving at a destination (the A-hah! moment). A Machine Learning model is likely used to estimate this time, which might not be something the investor even realized (the 2nd A-hah! moment). To ensure that the Machine Learning model is performing optimally, having a good MLOps system in place is crucial (the 3rd, and much weaker, A-hah! moment). We’ve found that most investors grasp the customer problem with a few real-world examples of products that they use on a daily basis, and how proper MLOps tools have a cascading positive impact on these products.


Making AI Less Emotional

Machine Learning and AI tend to be socially weighted topics from my experience. In other words, there is an unusually high proportion of individuals who have strong feelings (positive or negative) towards the technology to individuals who are indifferent. These strong feelings are rooted in popular culture with the notion that AI can only result in one of two outcomes: AI sending us into a future utopia of infinite wealth, energy and expansion, or in the counter scenario, AI leading to our ultimate demise.


My perspective is far less philosophical and much more pragmatic (and some might say boring…). We find it helps to take a very functional approach when discussing our company with investors to try and defuse any strong emotions about AI. The reality of today is that Machine Learning is currently being used by virtually every industry. To mitigate the risk of bad algorithms being deployed, repeatable and measurable processes must be implemented, and that’s what MLOps is all about. Quality control isn’t glamorous but it’s necessary for any type of product - whether it’s a physical smartphone, the software running on a smartphone, or the Machine Learning algorithm running in the software on a smartphone. Machine Learning is not the exception, and it shouldn’t be treated as such.


Pitch Events Produce Investors and Customers

One notable perk about fundraising for an MLOps company is that investor pitch events become venues for meeting investors AND customers. So many of today’s start-ups are looking to disrupt an industry with a unique application of Machine Learning and many of these companies are actively deciding what types of MLOps processes they should be implementing. For example at the latest pitch and networking event with VANTEC Angel Network we met almost as many potential customers as we did investors. It puts us in a great position to have investors watching us validate our business model in real-time at these events.