From Developer to Product Owner: The Fundamental Shifts from Generative AI

“We're not going to be software developers, we're going to be product owners,” Vince Passaro, Engineering Manager of Security at Stripe, recounts to Will Baxter in the latest episode of the Future of Threat Intelligence podcast. 

That realization, Passaro notes, was his pin drop moment. The moment when his mental model of what is possible in the engineering and security spaces drastically changed due to the availability of generative AI models.

In this episode of the podcast, Passaro and Will dig into what these models mean, how the center of gravity is shifting, and what is next for analysts, engineers, and security professionals. 

How Generative AI is Leading to a Fundamental Mental Shift

Generative AI allows for a certain amount of abstraction, notes Passaro. No longer do practitioners need to be involved in the “plumbing” of the systems being developed. 

This technology allows practitioners to no longer worry about finite technical details, for instance. Developers can now move away from concerns like specific Python versions or individual error debugging to focus more holistically on end goals and measurement criteria. 

“It's that abstraction back of thinking about what is the problem that I'm trying to solve?” says Passaro. “How do I measure that? What's the outcome?...That's effectively what you're doing. Okay. This is the end state, end goal. This is what I want it to be. And then work your way back.” 

This shift towards a higher level, holistic line of thought is a shift in commoditization. As generative AI can take over a lot of the skills traditionally used to manage complex infrastructure and the “plumbing” and “wiring” of these systems, the focus on humans shifts from being a product developer to a product owner. 

This empowers individuals who can see an overall role of the system being developed—from architecture to use case. 

How Generative AI is Impacting Productivity and Workflows

This shift towards the use of generative AI in development massively impacts productivity and workflows. Projects that could take years or decades to develop can now be prototypes in a matter of weeks by a single individual. 

Passaro notes that one of his team members built a complex product suite in two weeks. In traditional development cycles, Passaro estimates that the product would have required years or decades of work to reach the same point. 

“One of the engineers, Brian, he's just been on a rampage, effectively just redoing everything [with the help of generative AI],” Passaro notes. “Our whole analyst workflows, graphing functionality that we use on how we span accounts. You look at the product suite of what he put together in a couple weeks, years, I'd even probably argue decades or not possible. Yeah, just would never have gotten staffed.”

This shift enables work that would never have been possible otherwise. And with budgets stretched and the economic environment uncertain, this stretching of resources helps organizations and individuals flex their productivity while passing through potentially lean times. 

This compressed timescale is a significant advantage for “doers,” Passaro notes. AI acts as a force multiplier for analysts and engineers by allowing them to solve their own workflow problems. No longer do these teams need to halt their own workflows to wait for external developmental teams when they can work backwards, own a product, and use AI to build it to the exact specifications needed. 

What Are the Challenges and Risks Introduced by AI Workflows

While AI brings significant advantages and benefits, it does also pose its own, new sets of organizational and technical challenges. The main challenges introduced for developers and product owners are: 

• Production silos: As everyone becomes a product owner, and vibe coding enables personalized builds, organizations should ensure that products do not become siloed and inoperable with the broader enterprise architecture. 
• Data standardization: Data consistency can become challenging across AI-generated tools; however, AI systems may also help machines “work it out” and standardize the data. 
• Maintaining focus: As generative AI makes it easy to iterate on tools, Passaro warns of a “dopamine-fueled” loop of feature additions rather than focusing on the development of the core tool and the problem it is intended to address. 

In addition to the challenges posed to defenders, attackers are also using the same technologies. This also enables attackers to scale, and puts further pressure on defenders to outpace them. 

Key Perspectives and Advice for
Using Generative AI 

For individuals beginning to use generative AI for their workflows, Passaro shared a few key tricks to make the most of the tools. 

First, Passaro believes professionals should buy themselves a professional AI subscription. From there, developers should start “vibe coding” immediately. This experience will pay dividends when it comes time to actually build out new workflows or products. 

Second, Passaro encourages people to use AI systems as a teacher and a co-pilot. While people experiment with these tools and the outputs from them, they should try to soak up as much as they can from it. This means learning new technical parts of the workflow that they may not have been exposed to otherwise. 

Lastly, Passaro believes that organizations must establish frameworks and “laws” for AI-generated code and tools. This ensures that organizations maintain production standards. It can also help reduce production silos, data standardization issues, and maintain consistent workflows across a company. 

‍Listen to the full episode of the Future of Threat Intelligence Podcast with Vince Passaro, Engineering Manager of Security at Stripe, HERE.