At Cohesity, Eric Brown is dealing with two practical pressures: the tension between CFO and COO responsibilities, and the early impact of AI on how sales work gets done. From managing budgets and capital allocation to designing sales compensation and evaluating new tools, his role spans both financial oversight and day-to-day operations.
In this first of a two-part feature, Brown spoke with Frontier Enterprise about his path from finance to tech, how he handles the trade-offs between his dual roles, and why parts of the sales process, particularly lead qualification and handoffs, are starting to be automated as AI tools are adopted.
You were previously with Informatica when it went public. What do you remember during that time?
I joined Informatica in 2018. They were private at the time, having been privatised in 2015 by two private equity firms. I helped take the company public in October 2021. It’s interesting because we did a lot of the work on the IPO, including the SEC Form S-1, while working remotely. We were fortunate because collaboration tools like Zoom had advanced just in time when everyone had to leave the office during COVID. There was a small upside in that period in terms of having those tools available. We took the company public, and it’s interesting to now see it being acquired by Salesforce. It feels like a full circle moment.
I have a pretty deep technical background, and I’m probably closer to the tech than most CFOs. During my nearly five years at Informatica, that experience is relevant to what we are doing at Cohesity today, including helping a company manage a heterogeneous system and data environment, and going through the full end-to-end process of taking a company public and beyond. An IPO is a milestone for Cohesity. Having gone through it before and learned from a couple of mistakes along the way, I expect it to be easier the second time, if it happens.
How did you transition to tech?
Early in my career, I had a classic finance director background. I lived and worked in APAC for two and a half years, and then in Europe for another two and a half years, taking on larger and larger businesses as a finance director. As I was about to move into the next role, I decided to start a software company with a group of friends. That’s what I did in the 1990s. We called it a data mining company, but it was like a Gen 0.1 AI company. I’m a big believer in being a lifelong learner, and I’m certainly that.
You’ve been both COO and CFO. Is there inherent tension between those roles?
Yes, there is always tension. You have to understand where to create that tension point. For example, I may have to manage budgets and expenses. People always want more money. No one ever shows up and says, “I want to spend less.” Therefore, you have to be the arbiter of how to best deploy company expense and capital resources. Nothing ever changes there.
But the flip side is that you’re also operationally involved, like how you design a sales compensation or commission plan. That’s one of the most important decisions you make as an enterprise tech company with a direct sales force. I’ve sold software myself before. I’ve made cold calls. I’ve negotiated complex agreements end to end. So I know how hard it is to sell enterprise software. I understand every single step along the way, so I can design a sales incentive plan to produce the right behaviours.
Speaking of sales, will AI fundamentally disrupt that process?
I think certain roles will be disrupted by AI. We’ve adopted an AI-native tool called Outreach.io, which is a prospecting and outreach tool that automates a significant portion of what a typical inside sales or commercial sales rep would do. Think about when you hire people for lead generation or inside sales, or when they’re taking a curated lead from a Marketo system with a score above 80, or whatever your criteria is for a marketing qualified lead, and then passing it on as a sales accepted lead. That critical interface and handoff is typically handled by humans. What do those humans do? They run a pretty predictable process, end to end.
Now, what if you could script that process to take a lead, nurture it, and convert it into the pipeline using AI tools? With the appropriate scripting, tools, and data linkages, you could do it. For example, tools like this can capture conversations a sales rep has with a customer or prospect. They can join a Zoom session automatically, with permission, convert speech to text, run the output through an LLM, and generate insights, key value drivers, and follow-up questions from the call. They can also log into the CRM system and write the information back into Salesforce. That’s a big time saver.
Why do most AI pilots fail to reach production?
I think collectively, we’re flailing right now because things are changing so quickly. I used to think they were changing weekly, but they’re changing every day. The rate of change is very high, and we haven’t been at this rate of change for that long. It’s really only been within the last 12 months or so. What I tell my team is that they have to jump in and start using the tech to understand how it works through trial and error. Watch YouTube videos, ask the AI how to use the AI, and figure out how it works.
Very early on, around nine months ago, in my finance and IT teams, I said, “Look, we’ve hit an inflection point. Our professional lives have irreversibly changed, whether we like it or not.” I looked at the benchmark scores of the frontier model at the time, which was GPT-5.1, and the general consensus was that the model had reached about 140 IQ, which is generally considered genius level, across a wide range of subjects. You could access a free version, or a US$20 per month version, and effectively have access to an army of geniuses. At that point, I was using it for contract summaries, email summaries, and similar tasks. This ability to synthesise data was unbelievable. Back then, the ability to do research was also unbelievable. This was before we had more advanced agents.
For me, the proverbial light bulb went off, and I told my team that we’re all going to have to learn how to use this, because while their jobs are not going to be replaced directly by AI, they could be replaced by someone who knows how to use it.
