After two decades building and running technology companies, I wanted to approach Chapter 4 differently – not by forgetting what I’ve learned, but by questioning everything I thought I knew, because I believe with AI, the rules have changed.

That’s what Ascendius is about – and this post is the first in a series about building it. My goal is simple: build a technology company that generates more than $10 million in annual recurring revenue with fewer than ten full-time employees. This isn’t unheard of, but it won’t be easy, and I’m excited to give it a shot!

The Hypothesis

The hypothesis behind Ascendius – the parent company of TeamScore and what I hope will become a family of sibling products – comes down to three ideas.

1. AI can make talented individuals 5x more productive.
While I’m not sure whether it will be 2x or 5x or 10x, I’ve already seen this firsthand through building TeamScore. AI tools make it possible to plan, write, code, and market faster than ever before – with quality that’s not “good enough,” but genuinely impressive. If that compounding advantage continues, we’ll be able to ship world-class products faster and cheaper than previously possible.

2. Post-AI companies have an unfair advantage.
Starting fresh matters. Established companies have to wrestle with technical debt, organizational inertia, and the politics of change. When you start clean, you can design every workflow, system, and even incentive around AI from the beginning. That’s not just an efficiency gain – it’s structural leverage.

3. Solving the CAC crisis is critical.
Even as it’s become cheaper to build software, it’s become harder to get it in front of the users who need it. Inboxes are scorched earth where we hover over the “report spam” button. No one answers a call from a number they don’t already know. The ad duopoly of Google and Meta extracts every last dollar of marginal spend, while Apple’s 30% outrageous “tax” continues to impair innovation.

So despite the cost of creation dropping, the cost of acquisition keeps climbing. That’s a crisis. 

I don’t know how to solve it yet, but I think the answer lies in combining audience-first thinking, cross-selling across a product portfolio, and AI-driven marketing that’s genuinely helpful rather than spammy. It’s one of the puzzles I’m thinking the most about.

The Productivity Promise

There’s no doubt we’re in the upswing of the AI hype curve. Anyone who’s used AI to do their job knows that feeling of awe when they completed a task way faster than before. However, a recent MIT report also found that 95% of corporate AI pilots are failing. But just like when the internet first emerged 30 years ago, it is clear to anyone who’s used it that this technology is powerful and transformative. 

One of the reasons I think big companies are struggling is because they’re trying to eliminate all of a lower-level role before applying the technology further up the expertise stack. This made sense in the industrial era, where robots did rote, repetitive tasks, but in the post-industrial knowledge economy, AI doesn’t completely eliminate one type of job at a time. Instead, it reshapes every job it touches and often has a bigger impact at the non-routine, non-rote work higher up the experience stack. In my experience, it doubles the productivity of almost every knowledge-based role – including up to the CEO and Board.

That’s the unlock. You can use AI for the things you used to hire an analyst, a marketing agency, or even a strategy consultant to do. For a few dollars a month, and instantly. 

AI is an exoskeleton for talented, creative people – not a replacement for them.

The companies seeing results are the ones where people seek the unlock instead of fearing it. Where AI isn’t a threat, but a multiplier.

At Ascendius, I use AI as an active partner for multiple hours every single day. I use it to brainstorm strategy, design architecture, refactor code, and, of course write 3 or more blog posts a week. The productivity gain isn’t about speed alone – it’s about the quality and breadth of what one person can now achieve.

But what I also know first hand is that “vibing” doesn’t work. I’ve been as excited as anyone to see a prototype come to life before my eyes, and the first version of TeamScore was an MVP alpha built super fast. But whether your coding or writing a legal brief or a consulting report, vibing doesn’t work. As a technology product, vibed code is unmaintainable and often insecure. As a marketing and sales tool, set-and-forget AI tools do more harm to brands and products than they save in time. 

However, if you use AI as a multiplier instead of a replacement, it is transformative.

For example, with TeamScore I was able to build a powerful, multi-region product with two dozen connectors in a programming language I didn’t know on a back end I’d never used in <6 months. It is how I was able to create a 30 page go-to-market plan that should take over 3 months in under 3 weeks. It is how I’ve been able to do detailed analysis of data in a couple of days that would have taken a couple of weeks, and of course how I’ve been able to write this and all of my other blog posts over the last two weeks while doing everything else.

That’s the productivity promise – and it’s already real.

The Advantage of a Clean Start

Most established companies are trying to retrofit AI into organizational structures, processes, and policies built for a pre-AI world.

Those structures weren’t designed to resist change – they were designed to manage risk and maximize the consistency of people. But now, every role, policy, and workflow is a piece of friction resisting change whether actively or accidentally. When people evaluate AI through the lens of how to do their job rather than asking whether their job should exist, progress slows to a crawl.

As Upton Sinclair wrote back in 1934,

“It is difficult to get a man to understand something, when his salary depends upon his not understanding it.”

It’s not malice. It’s human nature. It hasn’t changed in the 90+ years since Sinclair wrote it, and it isn’t going to change any time soon. 

For technology companies, the problem runs even deeper than processes, bureaucracy and politics. It’s not just the people – it’s the code.

Joel Spolsky, the doyen of software engineering and founder of StackOverflow and Trello, wrote the canonical warning more than 25 years ago in “Things You Should Never Do, Part I” . Rule number one being never rewrite your software from scratch. It was true then, and mostly still is.

But now established technology companies face a paradox. To take full advantage of AI, they need to modernize their infrastructure, data models, and workflows. Yet for any company more than a few years old, doing so requires breaking Joel’s first rule.

While you can bolt AI onto an old codebase, you’re not going to see many benefits – at least not compared to AI-native tech companies.

All of this gets even harder when the company is owned by private equity – where the plan to flip a company in 3-5 years isn’t compatible with the timeline or investment required to take advantage of AI. The founders are gone, the MBAs are in charge, and while they’re good at doing acquisitions and pricing strategy, product and engineering innovation isn’t usually their sport.

That’s why starting fresh is such a competitive advantage. In addition to being able to harness powerful tech like Cursor or Claude Code because your code isn’t legacy spaghetti, there’s also no team defending the old system, no hierarchy to preserve, no compliance department standing in the way of experimentation.

And it’s why this era feels so exciting.

Many of my friends who’ve exited their companies are back at it again – not because they need to, but because it’s rare to get both the experience of having built before and the freedom of a blank slate.

In a world of massive change, that’s the sweetest combination there is.

Solving the CAC Crisis

While AI promises to accelerate the product engine of a tech company, customer acquisition cost (CAC) remains a massive choke pointw

For years, the cost to build a new software product has fallen. But the cost to find customers has continued to increase.

Google and Meta’s action-based advertising duopoly soaks up every incremental dollar of acquisition budget, while Apple’s evil 30% tax and self-preferencing stifles innovation. Old sales playbooks continue to see diminishing returns: the inbox is scorched earth with cold emails quickly getting the “report spam” click, no one answers calls from phone numbers they don’t already know anymore because of scammers.

At the same time, AI will make it even cheaper to build new products. The result is a flood of competition fighting for the same attention, in the same channels, with the same tools.

That’s the CAC crisis.

If this era is going to produce a new generation of durable software businesses, we’ll need new distribution models to match, and being 5x as efficient in your biggest cost – payroll – provides the ability to invest in a better product at a better price along with new go-to-market tactics. These could include audience-driven portfolios, value-based bundling, or deeply automated go-to-market loops that are personalized instead of pushy. There’s also promise in new players providing new paths to discovery – as long as we can keep the AI-slop at bay.

I don’t have the full answer yet. But it’s one of the most interesting challenges in modern entrepreneurship – and I’m thinking about it every day.

Let’s Go!

We’re living through the biggest change in technology since the internet went mainstream 30 years ago – and it might prove even more consequential than that.

For builders, it’s a once-in-a-generation opportunity to rethink everything – not just products, but companies themselves.

So that’s what I’m doing. And if you’re building too, I hope you’ll come along for the ride.