Introduction
We like to say that technology is a means to an end. With that in mind, the business of AI is the business of applying new technology to solve business problems.
As someone with an M.B.A. who loves the interplay of emerging tech and business fundamentals, I’ve been on my own AI journey alongside participating in the Perplexity AI Business Fellowship. My goal: learn and experiment to make the business of building great software better.
The program’s webinars have been fantastic because the practitioners sharing their experiences are living this evolution everyday. Three sessions in particular stood out for practical business insights that go way beyond the AI hype cycle.
Here’s what I learned from three leaders who are building, implementing, and investing in AI to solve real business problems.
Democratizing app development
Guest: Matt Palmer, Developer Relations at Replit
Business teams and non-engineers are no longer “locked out” of app creation. The traditional gatekeeping of software development is dissolving, and domain experts can directly translate their ideas into working solutions.
Matt Palmer’s fireside chat for the Perplexity AI Business Fellows demoed Replit and talked about how business teams and non-engineers can create functional applications without traditional coding skills.
Note: There is a limit to what you can build as a non-developer when complexity is a factor. And in business, complexity always becomes a factor. Read more about when low-code-no-code makes sense and when it doesn’t. And check out our blog and video resources for pragmatic approaches to agentic coding.
Disposable apps change everything
The concept of "disposable apps"—functional prototypes built in 30-60 minutes for one-off tasks—represents a paradigm shift. Instead of long development cycles, teams can now move from idea to MVP rapidly, creating apps that work rather than static mockups. This speed enables unprecedented experimentation and validation.
As our senior software engineers and senior product managers have discussed, the real value comes in treating these as true throwaway prototypes to accelerate the product software development lifecycle.
The "Vibe Coding" framework
Everyone loves a framework. Matt suggested AI-assisted development requires mastering five essential skills:
Thinking: Developing computational and procedural thinking to break down complex problems into manageable components.
Frameworks: Knowing what tools and libraries exist (React, Express, etc.) without having to implement them from scratch.
Checkpoints: Using version control to safely experiment and revert changes when needed.
Debugging: Embracing methodical troubleshooting when things break—which they will.
Context: Mastering the art of providing the right information to AI tools for optimal outcomes.
For a different approach to agentic coding with rapid prototyping followed by a product mindset stage, check out The Double Loop Model by Joé Dupuis.
The new reality
Matt also pointed to five key shifts in how we approach building:
- Mindset Evolution: From "learning to code" to "learning to build with AI"
- Rapid Validation: Testing ideas before committing resources
- Iterative Development: Embracing the experimental, non-linear nature of AI-assisted building
- Communication-Centric: Success depends more on communication skills than coding knowledge
- Functional Applications: These tools create real, deployable applications, not just demos
The democratization of app development through AI tools means who gets to build and how fast ideas can become reality. As the barriers to creation continue to fall, the most valuable skill is being able to think clearly about problems, communicate with AI tools, and maintain the curiosity to learn from success and failure.
What value does your business provide?
Guest: Eric Glyman, Co-founder & CEO of Ramp
Full disclosure: Test Double is a Ramp customer. We love it, but I’m not including highlights from this fireside chat just because we’re a customer.
Eric’s conversation was really interesting and focused on some great business themes that, while viewed through the lens of AI, apply to many other areas of business.
AI as a time multiplier for high-value work
Eric said AI should get rid of low-level tasks that pull skilled professionals away from their core expertise - using the example of world-class researchers doing expense reports instead of research. The focus is on getting time back for higher-value activities.
Going beyond traditional product categories
Eric encouraged companies to think beyond their original form factor. Using American Express as an example (from Pony Express to credit cards), he pointed out that successful companies identify their core value proposition (in AmEx’s case, "selling trust") rather than getting locked into specific products.
"Reasoning at scale"
Finance is fundamentally about reasoning—deciding who spends money, how, why, and under what conditions. As AI’s reasoning capabilities improve, it can increasingly handle these decision-making processes, moving from being a "nervous system" to becoming the "brain" of business operations.
Positioning as a productivity company
Rather than positioning as fintech, Ramp sees itself as fundamentally a productivity company that happens to work through financial tools. The core mission is making "dollars and hours go further"—optimizing both capital efficiency and time allocation.
Why positioning as more than an AI company matters
AI companies are due for a big shift in positioning: from tool providers to business solutions companies.
As with most new technologies, first there’s early adoption and then a huge rush as everyone gets into the space. This usually ends up feeling exciting but is often paired with people talking about “fill-in-the-blank” strategy and describing themselves as a “fill-in-the-blank” company.
If you re-play the past six months in your head, you will inevitably remember dozens of people talking about needing to create or implement an AI strategy, or describing their business as an AI-first such-and-such company.
As things level out, yes, smart leaders leverage the new hotness tech to enable the actual strategy. But they position their companies at the core around something that stands the test of time and tech: help solve real problems for real people. Ramp is doing just that.
Beyond the basic boosts to business
Guest: Roy Bahat, head of Bloomberg Beta
I love journalism, so it was great to hear from a leader with the VC arm of a media company talk about practical applications of AI in business.
AI adoption follows classic technology adoption patterns
Roy Bahat posited that AI is "new technology on steroids"—faster and more powerful than previous innovations. We’re in the trial-and-error phase, just like any other transformative technology.
Every company must become a tech company
Companies that don’t adopt a tech mindset will miss AI’s competitive advantages. The big shift happening now in business is the old school mindset of incorporating AI requiring corporate-wide initiatives to more freeform: individual employees having direct access to AI tools for personal productivity.
Looms vs. Slide Rules vs. Cranes
Another framework!
Roy shared this one for thinking about which applications of AI are most valuable to businesses.
Looms: Automate existing work (can replace workers)
Slide Rules: Enhance individual productivity (make people better at their jobs)
Cranes: Enable entirely new capabilities (allow organizations to do things previously impossible)
The most powerful applications of AI tools are "cranes"—enabling work that couldn’t be done before, not just automating existing tasks. I found this framework really useful for thinking about technology in business more broadly.
Tip for using AI in your workflow
Start small, build intuition through trial and error
Focus on the most annoying, least valuable parts of your day. AI isn’t magic—it’s a tool that requires learning and experimentation. Don’t apply AI to mission-critical functions without understanding its probabilistic, non-bulletproof nature.
As our Engagement Partner Dustin Tinney likes to say: would you trust this to build the software that’s used to fly a plane?
Future job creation driven by human demand, not AI supply
New jobs emerge from human needs (aging population needs healthcare, demand for more software to fix problems) rather than technological capabilities. The rise of "artisanal" work—jobs where “human provenance” matters—represents the only truly automation-proof category.
Test Double co-founder Justin Searls has some thoughts on how to tell if AI is going to take your job that align with this idea.
Transition challenges require societal response
Even if net job creation occurs, transition periods can be socially disruptive. Business leaders must consider broader societal implications and support systems during AI adoption phases.
Roy was right. Seriously, this is what gave birth to change management. People get squirrelly when there’s a lot of change, and leaders need to help them through it. Learning resources, growth time to spend learning new technology adoption, and encouraging sharing what you’re learning in the open.
Conclusion
These three conversations illuminate a common thread: AI isn't just another technology upgrade—it’s changing how we think about capability, value creation, and competitive advantage.
Whether it's democratizing app development, helping redefine business categories, or enabling entirely new forms of work, the companies that will thrive are those that embrace AI as a strategic multiplier rather than just an efficiency tool.
But all of that always comes back to business fundamentals.
The key insight across all three discussions is that success requires both tech fluency and business smarts. It's not enough to understand what AI can do; you need to understand what problems are worth solving and how to position your organization for the broader shifts ahead.
The Perplexity AI Business Fellowship continues to deliver these kinds of actionable insights, and I’m really enjoying it. For business leaders navigating the AI transformation, the program offers exactly what's needed: evidence-based perspectives from practitioners who are defining the future of AI in business, not just talking about it. Definitely recommend applying the next time applications are open.
Cathy Colliver is Marketing Director at Test Double, holds an M.B.A. degree, and has experience across branding, positioning, demand, and performance marketing at startups, mid-size companies, and enterprise across software consulting, insurance benefits consulting, automotive marketing, media, marketing solutions, and theatre.
Related Insights
Learn how to use external data with different naming conventions in Python with Pydantic. The guide covers different approaches and pitfalls using Pydantic's alias and alias generator features.
