From Zero AI Strategy to 80% of Marketing Tasks Automated

The Situation

Aviatrix is a multi-cloud networking and security company valued at $2 billion. When GPT-3.5 launched, their CEO Doug Merritt (the former CEO of Splunk) knew the company needed to move on AI. But like most enterprise organizations, Aviatrix had a gap: deeply technical products, no product marketing function, and no clear path from AI curiosity to AI execution.

Brian Sowards was brought in by the CEO to lead an AI transformation effort. Josh Cridlebaugh, then a product management leader running Aviatrix's observability suite, became his primary partner inside the organization.

"The environment at Aviatrix wasn't ready for a digital transformation," Josh recalls. "And Brian still came in and was able to bring in change. The work that was done. It set the tone for the embrace of AI throughout the corporation. That would not have been possible without what he did."

---

The Problem Nobody Could Name

Josh had identified a core business problem: nobody at Aviatrix knew how to talk about their own products. The products were deeply technical and solved real problems, but the external messaging didn't reflect that. He sat down with Doug Merritt directly and said, "We have a problem: we don't have a product marketing group."

Doug's response was to throw a product person at it. Josh stepped into a newly forming PMM group. But the real question wasn't about marketing. It was about what to build first, and how AI could accelerate the answer.

"We had this new world of AI starting to come out," Josh says. "And we were like, we need to start using those resources. That was around the time Brian came into the picture."

---

The Art of the Possible: Then Picking One Thing

What Brian brought to Aviatrix wasn't a slide deck or a training seminar. It was what Josh calls "a sense of the art of the possible with AI."

"It was important for us to first get that overarching picture of... there are so many things we can do with this," Josh explains. "And then there's something I've stolen from Brian that I've used throughout the last few years of my career: We can do so many things. Let's pick one. Let's generate a proper ROI. Let's validate and prove it, embed it and test it. And then from there, we'll earn the right to move to thing two, three, four, and five."

This became the operating philosophy. Instead of scattering energy across a dozen AI experiments (which Josh watched happen repeatedly after Brian left), the approach was surgical: find the highest-leverage problem, prove AI solves it, then expand.

---

What They Built First

The first project was deceptively ambitious. Josh pulled seven years of sales win data from Aviatrix into a spreadsheet and started using AI to reverse-engineer what actually caused deals to close.

"I started using AI to help me figure out what was actually working. When we looked at the sales (if there were enough notes), what was actually causing the sale? Who was the persona? Who really, at the end of the day, was our ICP?"

From that analysis, they built a system that could identify ideal customer profiles, find similar prospects based on LinkedIn activity and job history, and generate rich prospect profiles, complete with what each person cared about, what they'd recently posted, and what technologies they were running.

"I could give that to my SDRs and AEs and say, here's a write-up with their name, what they care about, what they've recently posted about, what technologies they're using, and it was all something we could roll out in less than three minutes. From ingestion of data to a target list."

When Josh demoed this to Doug Merritt, the CEO's response was immediate: "I want you to do this for the whole company."

---

From Champion to Director of AI

Josh's trajectory tells the story of what happens when AI transformation actually takes hold. He started in product management. Moved into PMM to fix an evangelism problem. Built AI-powered tools with Brian's guidance. And then was formally moved into a new role: Director of AI for Product Management, a position that didn't exist before the engagement.

"Brian shared with the leadership team that he felt I was the right person to move things forward," Josh says. "And I did that for about a year and a half, two years."

Brian describes the dynamic simply: "In an AI transformation effort, half the rocket's fuel is spent getting the first foot off the ground. That was me. And then the rest was Josh."

---

Enablement That Actually Scaled

One of the most telling outcomes was the internal enablement program Josh carried forward after Brian's engagement ended. They launched weekly AI office hours: 30-minute sessions that opened with 10 minutes of "here's what we learned this week" and closed with 20 minutes of open Q&A.

At a 300-person company, the first session drew 50 people. Then 70. Then 90. Then over 100, more than a third of the company, every single week.

"That was primarily because they wanted those first 10 minutes of what did these guys learn new about the tools they're testing this week," Josh says. The sessions also covered fundamentals: how to prompt properly, why AI isn't a Google search, why you need to give it context like you're talking to a peer.

The approach worked because it felt personal. "Transformation has to feel personal," Brian says. "It doesn't just happen at a business or systems level."

---

The Result

By the end of August 2025, Aviatrix's marketing team had achieved an 80% reduction in manual labor through AI automation. The story was covered by Business Insider, which reported that CMO Scott Leatherman's team now had each member running four or five dedicated LLMs for different functions, publishing six blogs a week and over a dozen social posts, while focusing their human energy on the 20% of work that required creativity and empathy.

"The foundational work we did was based off of what we learned from Brian: sequential prompts, proper AI architecture, and the thought process behind it," Josh says. "That's the reality."

---

What Made It Work

Josh distills the engagement into a clear framework:

Phase 1: The Art of the Possible. An outside perspective that sees opportunities the organization can't see on its own. Not a generic AI overview, but a deep, context-specific mapping of what AI can do for this business.

Phase 2: Pick One and Prove It. Resist the temptation to start 13 projects. Identify the single highest-ROI problem, build a working solution, and earn the right to expand.

Phase 3: Enablement and Evangelism. Create a repeating touchpoint where people can learn, ask questions, and see real results. Make AI adoption feel safe, accessible, and personal. Not threatening.

"The biggest lesson I learned," Josh says, "is that if you can do everything (if you can be Dr. Strange), the question is: what powers are you actually going to use?"

---

Josh's Advice to Leaders Considering AI Transformation

"People get model crazy. Being willing and able to build your system in a way that it's not tied to a single provider is incredibly important. Treat LLMs the same way you treated networking architecture with failover data centers, with that same level of resiliency. Even a subpar system is better than no system at all for a short period of time."

And on the human side: "If you don't learn this, someone who knows how to do this will take your job. That's the reality. But it's not as hard as you think. My dad is 65. He started a property management company five years ago, and in the last year we've injected AI into his tech stack. He's grown his business 3x. That's the real power: being able to take one person who's an expert in their field and multiply them."

---

Josh Cridlebaugh served as Director of AI for Product Management at Aviatrix, where he led the company-wide AI transformation that resulted in an 80% reduction in marketing manual labor. He previously ran product management for Aviatrix's observability suite and co-founded the company's product marketing group alongside CEO Doug Merritt.

Brian Sowards is the founder of Sowards.AI, where he helps companies move from AI curiosity to AI execution using the "Ship Two Things" method.