For all but a select few who jump in early, establish REAL value, and continuously deliver real innovation in how they leverage AI — most companies are trying to build a business solely atop a temporary rarity that will soon become a commodity.
As I’m sure it’s been for everyone over the last year or two, AI has been a huge focus for our team here at tension. Equally exciting, daunting and at times, especially early on, and always a little scary: when will it just replace me?
Things are moving at a crazy pace with AI, faster than any previous technology we’ve ever seen, and it’s really hard to keep up with everything. I mean, a couple weeks ago was different before Claude Sonnet and Claude Code were released, yesterday was different than it is now thanks to OpenAI’s 2025 Dev Day, and Thursday will likely be even more different than it is today thanks to Google’s Gemini @ Work live stream. It’s insanely exciting, and awe inspiring — but also exhausting and confusing. How the hell is anyone supposed to catch up?
If you’re feeling the same, you’re not alone.
Over the last few weeks, mostly to make sure I haven’t gone off the deep end, I’ve made it a point to talk to as many folks that I respect in the industry as I can about their personal experiences with and perspectives on AI, what it means to them, how they are and aren’t using it, what excites them, and what their greatest fears are around it. Many of those I’ve had the opportunity to chat with, across industries, verticals and roles, are still hesitant, doubtful, worried — or, on the other end of the spectrum jumping in feet first and hitting some pretty solid walls at a high rate of speed.
From what I’m seeing and hearing, there’s still significant confusion and ambiguity around how to effectively leverage AI, and in many cases — how to make money from it. While those two concerns may seem unique, they’re in fact the same: how can one not be left behind in this new technological wave — or, even more simply: how does one capitalize on AI to advance their company, personal productivity, and earnings?
Unfortunately, this is the simply wrong question.
Even with Claude, or AgentKit, or GPT Apps and their conversational OS, or whatever the hell Google announces in less than 48 hours — none of these mind blowing advancements will change much of anything except for the profit margins of Anthropic, OpenAI, and Google. It will not help companies be better businesses, it will not make bad ideas good, half assed attempts full fledged, and not make apps that solve the wrong problems, somehow solve the right ones.
With every new tech that comes across the horizon, there’s a gold rush to figure out how to monetize it as quickly as possible. Faster to market is more important than going to market with the right thing. This is not unique to AI, but it certainly is amplified with people using AI to vibe code an AI-based product, or agents for everything from email composition to interview analysis. The problem is, that for all but a select few who jump in early, establish REAL value, and continuously deliver real innovation in how they leverage AI — most companies are trying to build a business solely atop a temporary rarity that will soon become a commodity.
When everyone can offer what you offer because everyone has access to the same models, if you are not using them in new and interesting ways, to solve problems no one else is, you have no moat, you have no differentiation, and that means, you have no business.
Caveat: if you’re building your own models from scratch, from your own clean data, and do it well — that is an entirely different story, and a story for another day…
Unless you’re OpenAI, Anthropic, Google, or Perplexity amongst other frontier model developers, AI alone is NOT a business model. You can’t just repackage one of their models, add a little of your own zest to the mix, and expect to make enough profit to pay for the development of the tool itself. That’s a losing battle, and one any of the large players could decide to add your entire business model to their platforms as a very minor feature update and kill your business off in the bat of an eye. The AI version of getting “Sherlocked”. How do you think Zapier, n8n, Langflow, hell, Apple and Microsoft among others feel after today’s OpenAI keynote? I am sure there’s some late night meetings happening as I type this.
Startups and businesses who don’t build their own frontier models can’t think about AI today as a thing in unto itself, not as a product to sell, not as a feature or set of features to add to your existing product, not as a value add or a differentiator even.
We can only think of the application of AI today as a set of tools, accelerators and opportunity-creators that make three things possible:
To be honest, that’s all there is to it. Three things. Don’t underestimate them though, there’s gold in each of those mines.
We can accelerate the things we deeply understand with AI
The first is straightforward. There are a lot of unlocks in streamlining — it’s the foundational value proposition of Business Process Improvement — and it can be unlocked far more easily and feasibly today than ever. Offload rudimentary, repetitive or rules-driven tasks to AI. Have AI do preliminary analysis and have a qualified human in the loop to verify & validate. Leverage AI APIs to scour the web for supplementary information you don’t have resources or time to source and analyze, or analyze the data you already have in ways you previously couldn’t. Simplifying, augmenting, or extending capabilities to solve things you already solve in deeper, more meaningful ways.
We can do things we previously couldn’t with AI
The second is more of a ROI consideration for what are likely already backlog items if you have a solid product team. Things that may have cost to much to understand fully, or taken too long and too much money to develop a solution for, or patterns in customer behaviour that could previously not be discerned. Many of these things can be understood, analyzed, and solved by smart AI today — and when I say smart, I mean proper context, clean data, and strict guardrails. That being said, as everyone knows, we can do more of the things we deeply understand with AI
We can change the way we think about work with AI
In the third and the most advanced augmented capability, we can completely throw away the confines of how things were done, for how it could be done. I have been thinking about this for some time, and to be quite honest, it broke my brain for a solid 3 months. I pushed myself to look deeply at not the how, but the why of everything I thought I knew — and it was much, much harder than I thought it would be, even though I intuitively knew there was a new nut to crack. This one hurt, made my doubt myself, and took way longer than I would have expected from myself — but it hit me on my last day of vacation this summer (cause, of course it did). More, when I say “it”, I mean three distinct, separate concepts that together solved something conceptually for me that unearthed a new way of thinking about experiences for always-on, ever-churning and recalculating systems. That’s a topic for another day though.
These massive changes in the how and why can be really hard to identify, to pin down, let alone to work towards for most companies. It goes against nature. Habits, systems, best practices, routines — they make sense. They are something one can follow even when the brain quits. 1->2->3, A->B->C, if / else, then. However, those businesses that are open to, and willing to step back and look everything anew, with fresh eyes, with curiosity, with willful disregard for the status quo can open themselves to a new, tremendously profitable, world of possibilities.
At tension, we’re in the midst of reimagining what’s worked for us for 5+ years at the moment. We’ve tackled the first and second points, and now it’s a question of where do we gain the most value from human effort, and where do we need to gut check the AI… it’s completely changing how and how fast we do our discovery and strategy work — that too, is a story for another day though. ;)
That being said, that’s kind of all there is to AI today. It’s really that simple. All the hype, all the wonder, all the perceived magic, it all really just boils down to those three things. Better, enabled, or reimagined. Just like any other tech we’ve ever invented. AI isn’t the “thing”. It’s the tech that empowers you to solve the “thing” better, or that no one else has been able to yet, or completely rethink everything about the how and why. Simple right? Ha.
At the end of the day, AI can be a means to an end. Without an end, it’s just flashy, intriguing, and / or fun.
It’s not a business.
Starting something? Maybe you’re building a new product or service from the ground up, or revamping an existing one? We can help.
