The CSV That Beat Our App
We built a bookkeeping reconciliation tool for a client's CPA firm. Took us a few weeks. Custom-coded, tailored to their workflow, integrated with their existing systems. We were proud of it. Clean UI, solid logic, handled edge cases gracefully. The kind of product you point at and say, “this is why you hire a dev shop.”
Then a team member at the firm -- not even a developer, just someone who was curious -- took one of their reconciliation CSVs, dropped it into Claude, and typed something like “reconcile these transactions and flag mismatches.” Claude did it. Not perfectly, not with all the polish. But it got the core job done in about 30 seconds. The same core job our app was built around.
I sat with that for a while. Not because we did anything wrong -- the app was good work, and it had features the raw model couldn't match. But the gap between “good enough from a prompt” and “worth paying for from a vendor” had collapsed faster than I expected. The unique value window for what we built had shrunk from years to months. And for certain use cases, it had shrunk to zero.
That was the moment I stopped thinking about what we build and started thinking about what actually constitutes a moat in 2026. Because if a team member with no technical background can replicate the core function of your product with a foundation model and a CSV, you need to be honest about what your product really is.
Disposability Meter
Key Insight
If the core value of your software can be replicated by pasting data into a foundation model, you do not have a product -- you have a temporary convenience layer. The question is not whether AI will catch up to your features. It is whether you have anything beyond the features.
Software Is Now a Commodity
I have been building software professionally for years. I say this with no pleasure: the act of writing software is becoming a commodity. Not deprecated. Not worthless. But commodity. Like electricity, or bandwidth, or cloud compute. Essential, but not differentiating.
Claude Code is the proof. I use it every day. I watch it write entire features, debug complex systems, refactor codebases. It is not perfect, but it is improving on a curve that should concern anyone whose primary value proposition is “we write code.” A solo founder with Claude Code can ship what used to require a five-person team. A non-technical person with the right prompts can build a working prototype in a weekend. This is not a projection -- it is happening right now, in our shop, on our projects.
The micro-SaaS dream -- build a niche tool, charge $29/month, grow with content marketing -- is on life support. Not because the tools are bad. Most of them are excellent. But because the value they provide can be replicated by anyone with model access and domain knowledge. Without distribution, relationships, and integration depth, your beautifully-crafted SaaS product is a prompt away from obsolescence.
This does not mean software companies die. It means the moat has to be somewhere other than the software itself. The code is the table stakes. What you build around it -- the relationships, the contracts, the positioning -- that is the actual product now.
Unique Value Window by Year
Custom CRM
Someone already built yours last weekend with Claude Code
Internal Dashboard
Your intern just asked Claude to make one. It took 20 minutes.
Data Pipeline
Paste the schema into Claude. Done. Next problem.
SaaS Wrapper
The API vendor shipped your feature while you were reading this sentence.
Mobile App
There are 14 identical apps in the store. Yours is the 15th.
Key Insight
The unique value window for custom software is collapsing exponentially. In 2020, a well-built app had years of defensibility. In 2026, the same app might have weeks. Software is now a commodity input -- the moat must live somewhere else.
The Only Moat Left: Relationships
If software is a commodity, what is not? I keep coming back to the same answer: relationships. Real ones. The kind where someone picks up your call because they trust you, not because your product has a feature they need. The kind where switching to a competitor means breaking a partnership that took years to build.
Think about it this way. An AI model can write your entire application. It can design your database schema, generate your frontend, write your tests, deploy your infrastructure. But it cannot sit across the table from a CPA firm partner and negotiate an exclusive integration agreement. It cannot build the trust that makes someone choose you over a technically identical competitor. It cannot pick up the phone at 9 PM on a Friday when something breaks and make the client feel heard.
The companies that will survive this transition are the ones building relationship infrastructure. Integration agreements that create real switching costs. Partnerships that make you the gateway to an ecosystem. Positioning that means new entrants have to go through you, not around you. This is not some soft, feel-good business advice. This is hard strategy. The moat is contractual, relational, and positional.
We have started restructuring how QWave operates around this realization. We still build software -- we are good at it, and Claude Code makes us faster than ever. But the software is the delivery mechanism, not the product. The product is the relationship, the integration, the trust. The software is interchangeable. The relationship is not.
Competitive Moat Builder
Key Insight
An AI cannot build trust. It cannot shake hands, sit through difficult conversations, or earn loyalty over years of shared challenges. In a world where software is infinitely reproducible, the only defensible asset is the relationship. Build there.
The Capitalism Paradox
Here is where it gets uncomfortable. I just told you to build moats -- relationships, integration agreements, switching costs. That is good capitalism. Defend your position. Protect your margin. Survive.
But the technology is telling a different story. AI wants to be free. It wants to be open. The entire trajectory of the technology is toward universal access. Every model gets cheaper. Every capability gets more widely distributed. The walls we build keep getting shorter because the technology makes it easier for anyone to climb over them.
So which is it? Do we build moats or tear them down? Do we protect our position or share our tools with everyone? The honest answer is: both, and the tension between them is the defining challenge of this era.
Capitalism rewards moats. Humanity benefits from open access. And every founder, every employee, every creator alive right now has to figure out how to navigate that tension. You cannot ignore the market -- you will go bankrupt. But you also cannot ignore the direction of the technology -- you will become irrelevant. The path forward is narrow. You build defensible positions and you prepare for a world where those positions eventually erode. You capitalize now, and you give back as the playing field levels.
The Tension
Both are true. That's the paradox.
Key Insight
The paradox is real and there is no clean resolution. Build walls to survive today. Know that those walls will fall tomorrow. The companies that thrive will be the ones that build moats they are willing to eventually dissolve -- and that have something left when they do.
What Do We Do When Tech Does Everything?
This is the question I get most from people outside tech, and increasingly from people inside it. If AI can code, write, design, analyze, diagnose, plan, and build -- what do humans do? What is left for us?
I think the answer is more hopeful than most people expect. The roles do not disappear. They shift. When machines automated physical labor, we did not run out of work -- we moved up the stack to knowledge work. When AI automates knowledge work, we move up again. To creativity. To connection. To strategy. To the fundamentally human things that no model can replicate because they require a body, a history, and a soul.
The creator who has something to say will always matter more than the model that can say anything. The connector who has spent a decade building trust in an industry cannot be replaced by a chatbot with a LinkedIn scraper. The strategist who can hold contradictory ideas in tension and make a judgment call under uncertainty -- that is not a prompt-completion problem. That is wisdom, and wisdom takes a life to build.
The future is not humans vs. machines. It is humans doing what only humans can do, amplified by machines that handle everything else. The question is not whether there will be roles for people. The question is whether you are developing the skills that will matter in those roles.
Future Human Roles
Key Insight
AI does not eliminate human roles -- it compresses them to the irreducibly human core. Creativity, connection, judgment, embodied experience. The skills that survive are the ones that require being a person, not just processing information.
Where Do We Go From Here?
I wrote this piece because I needed to think out loud. The CPA firm incident shook something loose in my thinking, and I have spent weeks processing it. Here is where I landed.
The window is open right now. Not for building another SaaS app or another AI wrapper. The window is open for positioning yourself -- as a person, as a company, as a team -- in a way that survives the transition. That means investing in relationships. That means building integration depth instead of feature depth. That means being honest about which parts of your current work are replaceable and doubling down on the parts that are not.
It also means using the tools. Aggressively. If Claude Code can build your feature in a day instead of a week, that is five days you can spend on the things that actually compound -- partnerships, positioning, trust. The companies that treat AI as a threat are the ones standing still. The companies that treat it as a multiplier on their human strengths are the ones pulling ahead.
I do not know exactly what the world looks like in five years. Nobody does. But I know this: the people who will do well are the ones who are capitalizing now, building real relationships, developing irreplaceably human skills, and using every available tool to amplify what they are already good at. That is where we go from here. Not into obsolescence. Into the version of ourselves that the machines cannot touch.
Your Action Plan
Key Insight
The window for positioning is open now and it will not stay open. Use AI to accelerate execution. Invest the time it saves into relationships, strategy, and the irreplaceably human. That is not a platitude -- it is a survival plan.
The Bottom Line
Software is becoming a commodity. The moat is no longer what you build -- it is who you know, how deeply you are integrated, and what you offer that a model never will. The CPA firm incident was not a failure of our product. It was a signal that the game has changed. Capitalize now: use AI to move faster, build deeper relationships, secure integration agreements, and develop the human skills that compound over a lifetime. The future belongs to the people who understand that technology is the accelerant, not the destination.
Ready to figure out where you go from here?
We help companies navigate the transition from building software to building infrastructure that lasts.
Book a Discovery Call