As a startup founder, the most exciting signal you can get is when the market pull is greater than your market push. That’s when you know it’s time to scale.
I’m proud to announce that Moveworks has completed a $75M Series B financing round led by ICONIQ Capital, Kleiner Perkins, and Sapphire Ventures with participation from our existing investors Lightspeed Venture Partners, Bain Capital Ventures, and Comerica Bank. The round also included a personal investment from John W. Thompson, Partner at Lightspeed Venture Partners and Chairman of Microsoft.
The market pull is real. It’s our time to scale.
We believe that AI has the potential to resolve many common employee questions and requests automatically. We see this play out every day at our customers, where we’re resolving around 35% of IT issues autonomously. This financing round and the support from our new investors will enable us to really push the boundaries of what’s possible with AI.
Since coming out of stealth in April 2019, the interest in Moveworks has been overwhelming. We were featured on Forbes’ inaugural AI50 list of America’s most promising AI companies. And I can’t tell you the number of prospects and customers I’ve met with during this time, but I know it’s a lot. No matter how big this company gets, I still enjoy the thrill of meeting an IT leader for the first time and showing them a demo of Moveworks. Their reactions to seeing Moveworks in action are energizing—AI is magical when you get it right.
But what really sets our company apart are the stories from customers. I particularly enjoyed Broadcom’s recent blog about how Moveworks resolved 38% of their IT issues in their first year using our platform. Thanks for the kind words, Stanley Toh.
Getting AI to the point where it delivers real-world results—like what we’ve achieved at Broadcom—isn't easy. In fact, most companies are struggling to make it work. So I thought I’d share a few insights on how we make AI work at Moveworks.
Keeping the hiring bar high
Building AI requires a very different talent pool to traditional technologies: ML modelers, ML platform engineers, data scientists, data evaluators, hardware (GPU) specialists, and others. These people are well educated, highly skilled, and in high demand. You have to build a company that can attract and retain this kind of talent. To do that you have to vigorously stick to your rubric and hiring criteria, rather than succumb to the pressure of needing to grow the team by hiring someone who doesn’t meet your high standards. It’s the only way we can create the high-performance culture that propels our product forward.
Machine learning news headlines are dominated by advances documented in research papers. While many of these breakthroughs deserve the recognition they get, there is a huge difference between building models that work in controlled lab experiments (with clean data) and building models that work in real-time, live production environments with real customer data. Machine learning and natural language understanding are such fast-moving fields that staying ahead of the pack requires a culture of discovering new research, testing it, evolving it, and then productionizing it. This is a continuous cycle, not a one-off event.
Operationalizing machine learning
Our production environments are typically running thousands of machine learning models at any given time. But models are not like normal code. They drift, they get overfit, they degrade in performance. So building an automated data and model pipeline that can monitor performance, then retrain and redeploy models on the fly is a necessity for us to operate at scale. The reality is that there are no industry-standard frameworks, tools, or processes for doing this. You have to build it yourself. So you need innovative engineers at every layer of the stack.
General-purpose AI is still a long way off. Our success stems from our stubborn desire to stay focused on the IT support domain, and only that domain (at least for now). This focus has enabled us to become experts at applying AI to solve employees’ IT support issues. We have a huge catalog of labeled data, highly specialized language models that understand the nuances of enterprise IT jargon, a deep understanding of the processes that support common IT support use cases, and dozens of integrations into core enterprise applications. Staying focused has enabled us to reach the high levels of performance our customers benefit from every day.
2019 has been an incredible year for me, my co-founders, and the whole Moveworks family. We all know that we’re in the early stages of building a really special company. We’re truly excited to continue this journey in partnership with our investors.
To our customers: thank you for your support, you’ve been central to our success.
To the Moveworks team: thank you for your passion and relentless dedication. You are one of the best teams in Silicon Valley.
The journey continues.