By Dave Taddei 

The economics of software are changing quickly, creating new choices for growth companies and their boards. For years, companies have relied on generic SaaS platforms because building custom software was simply too expensive. In the process, businesses often adapted their workflows to fit what I call artificial constraints – limits set by the software they were using, not by the business. 

AI is changing that equation.  

Boards now have an opportunity to ask which processes should continue to rely on existing platforms, which could be redesigned, and where a more tailored approach could create strategic value. 

That makes AI governance part of a broader conversation about how the business creates value. Rather than adding another layer of oversight, boards and management teams should revisit the processes and governance practices they already have by asking:   

  • Where has technology shaped the way the company works?  
  • Which constraints no longer need to exist?  
  • And what could the business do differently now? 

For boards, those questions provide a practical starting point for evaluating AI investments and determining where AI can contribute to growth, efficiency and differentiation. 

 

Start With Business Value 

One of the biggest misconceptions I see is letting the technology lead. There is a lot of pressure right now to “do something with AI,” and that can quickly turn into getting a project out there without spending enough time on what the business is actually trying to accomplish. 

I tend to look at it from the business value first. What are we trying to accomplish? Where is there an opportunity to do something differently or remove a constraint? AI may be a useful tool in getting there, but the outcome gives you the starting point. 

I have long looked at technology spending through two lenses. Some investments create strategic assets that deepen what is unique or differentiated about the business. Others are commodity tools that are primarily a matter of cost management. AI makes that distinction even more interesting because companies have more options for what they can create themselves. 

Reimagine the Process 

Before approving a significant AI investment, I would ask management to show me the reimagined process. 

What artificial constraints existed before? Which ones can now be removed? What repetitive tasks no longer need to consume people’s time? 

That approach also helps prevent technology from leading the strategy. 

I saw this work well with Hopeworks, a nonprofit I advise. A few years ago, its leadership recognized that AI would significantly change the world, but they did not know exactly how. 

So, they started with some training for a small group of people in the organization. Those early learners became ambassadors who could help others get more comfortable with the technology. From there, they worked on smaller projects tied to specific outcomes.  

A few years later, that experience has spread much more broadly. The staff is AI augmented, as are the people going through Hopeworks’ job training programs. Graduates who embraced these tools are now competing successfully for roles against candidates who were taught to fear AI. 

I think there is a lot to take from that progression. The lesson for growth companies is straightforward: start small, learn, and build institutional competency. 

 

Avoid Deployment Without Enablement 

I’ve seen plenty of companies introduce an AI-augmented system, put it in front of the workforce, and expect people to know what to do with it. The training and learning piece of the puzzle can easily be overlooked. 

If someone has never used these systems or learned how to prompt effectively, it’s understandable that there may be some resistance. Giving people the opportunity to learn how the tools work and become comfortable using them is an important part of adoption. 

I think the same principle applies at the board level. I see boards trying to govern AI when some of the people involved have very little firsthand experience using it. Even relatively simple experimentation can help, like using AI to do some research or analyze a paper. Get a sense of what it does well, where it struggles, and how you might use it in your own work.  

The governance conversation also needs to extend beyond the technology team. Line-of-business leaders are especially important because they know the workflows that create value. They also know the constraints and workarounds their teams deal with every day. 

I would also recommend that the people closest to the customer be involved, whether that’s sales, customer success or another commercial leader. The customer experience is an area where there can be real opportunity to differentiate, and that perspective needs to be part of the conversation. 

Match Governance to the Risk 

Boards can encourage experimentation while maintaining appropriate oversight by considering what I call the “blast radius,” or the potential impact if an AI initiative doesn’t work as intended. 

A small internal system that can easily be shut down should need fewer controls than something touching customers, sensitive information or critical operations. The greater the potential impact, the more visibility and oversight the board may need. This gives companies room to start small and learn while applying greater scrutiny where the consequences are higher. There are smart ways to use any new technology, but non-use of AI is not one of them. 

That visibility becomes particularly important as AI takes on a larger role in software development. AI is already writing a significant amount of code, yet many organizations have little independent insight into the health and quality of what is being produced. . And the industry’s default answer — asking another AI to check the AI’s work — doesn’t close that gap. No board would accept management auditing its own financial statements; the same independence principle applies to verifying AI-generated code. 

For growth companies, a practical approach could include a quarterly AI scorecard built like any other business scorecard. 

Track the metrics that matter for the company’s current stage of AI adoption. That could include progress on key projects, system health, code intelligence or security findings. The weighting should reflect the company’s strategy and risk profile. Each metric should carry a baseline and a next goal level — what the board is really governing is the trend. 

 Make the Invisible Asset Visible  

In one independent code analysis I was involved with, 99 percent of the code was found to be AI-authored. It looked fine on the surface, but a closer review uncovered dozens of security findings (including live credentials committed directly into configuration files), along with quality and performance issues. In my opinion, that experience reinforced just how difficult it can be to assess risk based only on what appears to be working. 

Boards already use KPIs to understand the health of the business. I think we’ll increasingly see the same idea applied to code, giving boards greater visibility into the health, security and quality of  what may be one of the largest assets not on the balance sheet. That kind of visibility can help companies determine where additional oversight is warranted while still giving teams room to experiment and learn. 

Build a Learning System 

Looking three to five years ahead, I believe a significant divide will emerge between people and organizations that learn to work with AI and those that resist it. We went through a digital divide years ago; the next one will be an AI divide. 

Successful companies will build organizations that continuously learn and develop compounding institutional competency in AI. That requires experimentation, training and support from the board. 

And if I could give directors one question to ask management before approving the next AI initiative, it would be this: Show me the reimagined process and which constraints you removed. 

If management cannot answer that question clearly, the organization may need to rethink why it is pursuing the initiative in the first place. 

Author Bio 

Dave Taddei is an AI strategist, board advisor and venture creator with more than 30 years of experience across the technology and services sectors. He works with early-stage and growth companies to apply technology to practical business challenges, with a particular focus on generative AI, machine learning, and advanced analytics. 

 He is a partner at EMG.ai, an AI venture operating company, and serves on the board of The Code Registry, a code intelligence and code governance platform, as a Technology Board Advisor to Hopeworks, and as an advisor with Executive Leaders for Advisory Boards (ELAB). Previously, Dave was SVP of Global AI & Data Analytics Strategy at AllCloud, where he helped establish the company’s North American Cloud Data & AI Practice. He also founded and led data and AI professional services firm Integress as CEO.