Stephen Straus, CEO and Co-Founder of KUNGFU.AI – Interview Series

Stephen Straus, CEO and Co-Founder of KUNGFU.AI, is an entrepreneur, investor, and technology executive with decades of experience building and advising companies. Before launching KUNGFU.AI in 2017, he served as a General Partner at Austin Ventures and founded or co-founded several companies, including mbue, which was acquired by Bluebeam. Straus is also a longtime Partner and Mentor at Capital Factory and serves on the Engineering Advisory Board at the University of Texas at Austin’s Cockrell School of Engineering. Beyond technology and investing, he co-founded and chairs Impact Dyslexia and founded the Startup Diversity and Inclusion Pledge, reflecting a broader focus on how technology, entrepreneurship, and organizational leadership can create positive social impact.

KUNGFU.AI is an Austin-based management consulting and engineering firm focused exclusively on artificial intelligence. Founded in 2017, the company combines AI strategy with engineering to help organizations move from experimentation to production-ready AI systems, with work spanning areas such as agentic AI, machine learning, computer vision, AI governance, and human-machine teaming. Its approach emphasizes designing AI around how organizations and their people actually operate, rather than treating AI as a standalone technology deployment. KUNGFU.AI works across industries including healthcare, government, private equity, retail, finance, and insurance, and reports that 96% of its engagements reach production.

You spent years as a venture capitalist at Austin Ventures through the dot-com boom and bust, and later saw parallels between the rise of the internet and what was beginning to happen with AI. What convinced you the opportunity was significant enough to ultimately co-found KUNGFU.AI, and what did you see at the time that many business leaders were still missing?

I think it was pattern matching from having a front row seat to the internet revolution during the dot-com era that led me to co-found the company six years before ChatGPT came out. I can’t really explain it, but I got real conviction that AI was going to be the next big technical transition that the world would go through, and like the Internet, it would change the basis of competition.

Personally, in my life, I’ve seen multiple transformative technical transitions, starting when I was 11. I saw an Apple I, and though I didn’t know it at the time, it represented the start of the personal computer revolution.

At least from a services point of view, very few people saw the same thing that my co-founders and I did. Prior to ChatGPT, it was very hard to find other services firms focusing on AI, and now it’s almost impossible to find a services firm that doesn’t say they offer AI services.

KUNGFU.AI was founded years before generative AI became a boardroom priority. How has your definition of an “AI-first” company changed as the technology has evolved from traditional machine learning to generative and increasingly agentic AI?

When we started the company, we really recognized that we were pioneering what it meant to be an AI professional services firm. But it wasn’t until ChatGPT came along that we realized that we also had the opportunity to pioneer what it meant to be an AI-enabled professional services firm.

We had done some limited internal automation prior to ChatGPT, but it was really the advent of LLMs and now agentic AI that has enabled us to become an AI-enabled professional services firm. I think we are making significant strides towards that. We grew revenue over 80% last year, and we basically kept our head count flat. In services, both that growth rate and that uncoupling of the headcount growth is, I think, really remarkable and represents what’s possible in services in this new age of

You’ve argued that companies make a mistake when they treat AI like another software implementation. What does it actually look like for an organization to use AI to rethink what the company can be, rather than simply making its existing processes more efficient?

AI is different from traditional software on a variety of dimensions. This chart is from my new book, Wayfinding: A Business Fable About Leading in the Age of AI.

How AI is Different from Traditional Software

Beyond these specifics and more strategically, I think that most companies are making a really big mistake, and that is that they are trying to figure out how to use AI to automate their business. Why is that a mistake? Almost all those same people would agree that AI is changing everything. And if it is changing everything, that includes changing the basis of competition in your industry.

To me, it makes no sense to automate your business if your industry is going to be disrupted. That’s very much like saying it’s a good idea to automate Blockbuster Video when it’s going to be upended by Netflix.

It’s not that I’m against automation, but I think that every company’s first, second, and third priority should be to figure out how you can be the disruptor in your industry. Because if you are not focused on that, then by definition, you won’t be the disruptor and will likely be disrupted, you will be disrupted. So that should be the priority and only focusing on automation could be an existential risk for your organization.

KUNGFU.AI has now worked across more than 120 engagements spanning over 30 industries. From those deployments, what separates organizations that turn AI experimentation into lasting competitive advantage from those that remain stuck running pilots? 

There are really key differences that separate organizations that turn AI experimentation into lasting competitive advantage.  I think it starts from both the psychological profile of the leader and the level of ambition of the goals that they set out.

Let’s start with the psychological profile of the senior most leaders at the companies, whether that’s the CEO, but also the board and the C-suite. This is very simplistic, but I see two primary archetypes of leaders.

On the positive side, I see people who approach this massive challenge of leading in this new age of AI with the willingness to be vulnerable enough to say: This is completely new. We don’t know what we’re doing nearly enough to be able to make all the decisions we have to make and all the investments we have to make. And we are going to proceed forward knowing we’re going to make mistakes, and be comfortable with that. We’re going to learn from those mistakes, and we’re going to surround ourselves with experts as much as possible.”

That contrasts with people who, at least on the outside, go into this with the attitude of saying  I know what I’m doing, and with confidence that, for almost everybody, is unfounded, and therefore it’s more like hubris.

Beyond that, the ones who are most successful are the ones who realize that AI is going to change the basis of competition, and they are looking for AI-enabled business models built on proprietary data aimed at fracture points at the structure of their industry to be able to change that industry and then be a leader in it.

As AI agents become capable of executing increasingly complex workflows and making decisions with less human involvement, how do you expect the structure of companies, management layers, and individual roles to change?

All of this will change and is changing dramatically. We have a whole framework around human-machine teaming, which is taking research that is 50 years old and applying it to this new age of AI. The opportunity is not to just automate work or put AI tools in front of people, but to redesign the work itself around teams that include both human and machine intelligence.

That means the role of people will change as well. Increasingly, people will move from executing work to designing the systems that make the work happen, while bringing the judgment, oversight and understanding of context that the technology still requires.

It’s a very complex question. It’s happening fast. And it’s really important to be incredibly intentional about your people’s roles as they increasingly work with autonomous agents.

Companies now have an enormous selection of foundation models, AI platforms, agents, and third-party tools available to them. Where should enterprises rely on external technology, and which AI capabilities or data advantages do they ultimately need to own themselves?

With all the tools that are coming out, the intelligence portion of AI is rapidly being commoditized and I can’t emphasize enough the importance of having your own data assets.

There’s an expression that says the best day to plant a tree is 40 years ago. The next best day is today. What this means in relation to data is that if you are not collecting and archiving as much data as you can from your business, you should begin that process today, even if you are not going to do anything else as it relates to heading down the path towards an AI centric future.

Another way I think about this question is that there is a whole range of tools in the AI toolbox and the tools that people now tend to call traditional AI or machine learning models aren’t typically available off the shelf. And if they are, they are not necessarily going to get you the kind of accuracy you might need to get real business value out of them, and/or might not be able to utilize your proprietary data.

So I absolutely think it makes sense to rely on external technology. For many applications, the most strategic are likely to be ones that are bespoke solutions built on your proprietary data.

You’ve emphasized the importance of board-level AI governance. What questions should boards be asking management today to determine whether their AI strategy is creating genuine strategic value rather than simply producing incremental productivity gains?

Boards have a new set of challenges with AI. They have their traditional role of oversight and governance, which often is focused on downside protection and risk mitigation. But if they don’t balance that with encouragement of risk taking and trying new things and providing the resources to enable those, by definition the company will be left behind in this new age of AI. So finding the right balance between risk enablement and mitigation is key.

I also think it’s important for the board to be pushing management towards AI enabled business models that can change the basis of competition in their industry that are built on proprietary data, because if they are not focused on that, then the incremental productivity gains are going to be for naught when their business is disrupted.

In Wayfinding, you and Ben Szuhaj tells the story of a CEO who has 90 days to develop an AI strategy or risk being replaced. Why did you choose a business fable rather than a traditional management book to explore how leaders should respond to AI-driven disruption?

I’m a big fan of the business fable and authors like Patrick Lencioni, a legend of the genre. We chose to write this book as a fable because it helped us explore nuanced topics like the psychology of leaders to be able to be successful in this new age in an easy and approachable way that would have been much harder in a traditional business book.

We think that the book has really achieved something that we’re quite excited about, which is to teach really substantive and relevant material and do it in a way that is fast, fun and engaging for busy leaders. The learnings in it, including the 11 original frameworks, are things that readers can bring into work and put to use immediately.

A central theme of Wayfinding is making consequential decisions when technology is evolving rapidly and certainty is impossible. What framework should executives use to decide when they have enough information to act, particularly when waiting for greater clarity may itself create competitive risk?

One of the frameworks that we developed that we share in the book is, itself, called Wayfinding. It’s an adaptation of Agile for AI. It is a hypothesis-driven, structured process of exploring data and validating the feasibility of a solution before committing to a full production build. Its purpose is to answer whether an AI solution is both worth building, and will it work. It is a key reason that our firm has a 95%+ success rate for taking proofs-of-concept to production that have delivered real, measurable business value, which we have done over 50 times for our clients.

Putting your venture capitalist hat back on, when you look at today’s generation of AI companies, what signals tell you that a startup is building a durable competitive advantage rather than a product that could quickly become a feature of a foundation model or larger platform?

I really have no idea and am, frankly, very glad I’m not in that business today. I have a lot of empathy for today’s venture investors. They have to be taking big, bold bets if they want to achieve top tier returns for their investors. But because the technology is moving so fast and in quite an unpredictable way, it has got to be very hard to discern long term winners from losers.

I emphasize long term because unless a company gets lucky and hits on an important opportunity and gets taken out early, there are going to be a lot of companies that look like they have the opportunity to be successful, but end up not being successful because of the way the market ends up evolving, which again, is very hard to really know. Bottom line: like in the dot-com era, there are going to be some really big winners as well as a lot of money lost.

Thank you for the great interview, readers who wish to learn more should visit KUNGFU.AI.