AI Adoption in Action: How Revaia’ Portfolio CPOs & CTOs Navigate Transformation
When Revaia brought together product and technology leaders from across its portfolio for a workshop on AI adoption, the discussion quickly moved beyond theoretical possibilities to the messy realities of transformation.
The October session, part of Revaia's ongoing effort to foster peer learning across its portfolio, revealed both the tremendous potential and significant challenges of integrating AI into product development workflows.
Traditionally, the roles of Chief Product Officer and Chief Technology Officer have had a clear separation in terms of missions and duties. But AI is blurring those lines, changing the way product teams are building products, and the way tech teams are developing software and infrastructure. Both are evolving at a very fast pace and must rethink the connection between the two functions.
While there are a wide range of available resources, we find one of the best ways to learn is still through trusted colleagues. The diverse group convened by Revaia represented companies ranging from $5 million to over $70 million in ARR. They shared candid insights about their AI journeys in terms of how they are trying to find the right balance between testing everything and putting new ideas in production to create significant impact.
The discussion was guided by leaders from two of our newest portfolio companies, Definely, a legal AI platform that helps legal teams draft, review, and manage contracts more efficiently, and Ampeco, an EV charging management software platform, who presented detailed case studies that illuminated the path forward for others grappling with similar challenges.
Beyond Hype: Real Implementation Stories
While the sea change caused by AI has understandably provoked some anxiety, in general, we’ve found the mood among our portfolio companies to be one of excitement. After all, by their nature, these entrepreneurs and innovators are excited to try new things and see where they can take them.
At the same time, there is a sense of urgency given how fast all markets are moving now. In that respect, Bruce Elliott, Chief Product Officer at Definely, opened with a pragmatic view of AI adoption.
"It's about competitive advantage," he explained. "If we're not using AI, it's a competitive disadvantage."
This straightforward assessment set the tone for a discussion grounded in business reality rather than technological enthusiasm.
Definely's approach focuses on three key objectives:
- Removing friction in the product development lifecycle
- Ensuring better customer alignment
- Empowering teams to be more productive and self-reliant
The company integrated AI tools such as Cursor, Claude Code, and Figma into their design and prototyping workflows, achieving 30-40% efficiency gains for designers.
The demonstration Bruce provided was particularly illuminating. Within minutes, he showed how AI could generate multiple design variations and create functional prototypes based on simple prompts. An editing tool feature that might have taken hours to mock up traditionally was created in real-time during the session.
The Secret Sauce: Context Engineering
Alexander Alexiev, CTO of Ampeco, presented an even more radical transformation. His team achieved what many would consider impossible: complete automation of their software delivery process in just two weeks.
Ampeco's breakthrough came from recognizing that the key to AI success wasn't just about selecting the right tools. It was about “context engineering.” As Alexander defined it, context engineering is the practice of carefully crafting the information and instructions provided to an AI system to get consistent, high-quality outputs.
In Ampeco's case, Alexander and his team meticulously documented every aspect of their engineering process in markdown files, creating detailed prompts that explained each step from planning to implementation, testing, and deployment.
Think of it like the difference between telling someone to "make dinner" versus providing them with your kitchen layout, dietary restrictions, available ingredients, preferred cooking methods, and step-by-step recipes. The more relevant context you provide, the better the outcome.
"We just sat down and wrote down prompts and context for the LLM, basically the same way that I would explain our process to an engineer really, really thoroughly," Alexander said. This comprehensive documentation became the foundation for their AI agents' effectiveness.
The difference between tools mattered less than how they were configured. While Ampeco had experimented with various AI coding assistants, including GitHub Copilot and Windsurf, they found that Claude Code's ability to work autonomously for extended periods, ranging from 10 to 15 minutes without intervention, was crucial for achieving true automation.
The Human Challenge: Managing the Transition
To introduce the new AI-powered workflow to Ampeco's 50+ engineers, Alexander organized intensive, in-person workshops to demonstrate the powerful automation results achieved with Claude Code. These full-day sessions covered everything from LLM basics to hands-on practice with the new tools. The key message: engineers were no longer primarily coders but context engineers, responsible for guiding AI to produce useful outputs consistently.
To accelerate momentum around adoption, each company continued to deploy additional strategies.
Definely's Approach:
- Implemented an "AI-fluent" requirement for all new hires in design, product management, and engineering roles
- Maintained selectivity in showing AI-generated prototypes to customers to avoid "prototype fatigue"
- Used AI-generated prototypes primarily for medium to large-scale feature requests
Ampeco's Methods:
- Instituted weekly team meetings where engineers shared AI success and failure stories
- Implemented "small-scale post-mortems" for any story that wasn't truly complete, reinforcing ownership and accountability
- Emphasized that improving the AI tools should take priority over immediate task completion
These strategies addressed a crucial insight: AI adoption isn't just a technical challenge. It's a cultural transformation that requires deliberate change management.
Even though AI is powerful, you need to rethink processes and make sure people are accountable for the results. Alexander stressed that it’s essential to keep emphasizing the importance of this new priority, even if the benefits are not immediately obvious because all the little improvements made each day stack up over time and eventually will drive big gains in productivity.
Key Lessons for AI Implementation
Several critical lessons emerged from the discussion:
1. AI Is Not a Silver Bullet: Alexander shared a cautionary tale about a colleague who tried to use AI to search through a large, structured dataset. The approach failed because it was the wrong tool for the job. "What he actually needed was a program that would actually search a database," he noted. The lesson: AI excels at certain tasks but isn't universal.
2. Experimentation Must Be Welcomed: Both companies emphasized creating psychological safety for experimentation. Engineers need permission to "waste time" trying different approaches, even with customers waiting for features.
3. Non-Deterministic Outputs Require New Thinking: Developers accustomed to predictable, testable code must adapt to AI's variability. The same prompt can produce different results, challenging traditional software development paradigms.
4. Ownership Remains Human: Despite AI's capabilities, human accountability cannot be delegated. Both companies had to reinforce that AI is a tool, not an accountant.
5. Process Transformation Is Essential: AI enables fundamental workflow changes. Ampeco discovered they could create AI agents that function as both product manager and developer, enabling entirely new review processes that wouldn't be possible with human resources alone.
Looking Ahead: The Evolving Landscape
The rapid pace of change in AI tools emerged as both opportunity and challenge. During Ampeco's three-week workshop series, major model updates were released, requiring real-time adjustments to training materials. This volatility led to important advice about avoiding over-attachment to specific tools.
"Don't get too attached to your tools," Alexander said, noting how quickly companies leapfrog each other in the AI space. The focus should be on developing AI fluency and context engineering skills that transfer across platforms.
Despite the challenges, both companies remain committed to their AI transformations. Ampeco hasn't yet seen dramatic productivity increases because the learning curve is currently offsetting efficiency gains. But Alexander expects significant acceleration as skills develop and context improvements compound.
Fostering Collective Learning
We should note that AI hasn’t just changed the workflow for our portfolio companies. It has also changed the way we advise the portfolio companies and support their growth efforts, as well as the type of questions we ask during the due diligence process.
For instance, while we always support a company when they make C-suite hires, with a recent company, we pushed to make sure candidates were AI-native.
The workshop is another example. It’s one thing to read a post by a stranger on LinkedIn. But hearing from people you trust in a safe space, sharing candid successes and failures, enables the kind of insights that might take individual companies months or years to develop independently.
As AI continues to reshape product development, the experiences shared in this session provide a valuable roadmap for other organizations. The path forward requires technical capability, certainly, but success ultimately depends on managing human adaptation to radically new ways of working.
In a landscape where AI capabilities evolve weekly and best practices remain unwritten, this collaborative learning model is essential. At Revaia, we increasingly believe that part of our mission is to facilitate that learning. Because the companies that will win will be those who can adapt, learn, and transform the fastest.