Services
AI Solutions
I’m all about digital growth - for the past 10 years I’ve been working on digital innovations and building dozens of software products. I know each stage of the cycle like the back of my hand. As the CEO at Appunite I coordinate creating exceptional products with entrepreneurs and companies. The combination of my sales and tech experience makes me capable of understanding our clients needs and challenges, and it has helped companies from a variety of industries grow digital. My main goal is always to enable our client to focus on their industry transformation and the core of the business, while Appunite is responsible for building and delivering the products.
A fundamental shift is underway in Custom Software Development industry. For two decades, the economics of software have driven mid-sized companies toward purchasing off-the-shelf solutions. Custom development was expensive, unpredictable, and risky. The conventional wisdom was clear: unless you're a tech giant with unlimited resources, you buy—you don't build.
That calculus is breaking.
AI-assisted engineering isn't just making developers more productive. It's fundamentally restructuring the cost curves that have governed software development for decades. If you were a mid-sized company, the logic was simple: you bought software unless your business was so unique that custom development was unavoidable. And when you did build, you accepted the inevitable: long timelines, unpredictable costs, and the constant fear that the project might collapse under its own complexity.
In 1865, economist William Stanley Jevons observed something counterintuitive about coal consumption in Britain. When James Watt introduced more efficient steam engines, everyone expected coal demand to drop. Instead, it soared. The increased efficiency made steam engines economically viable for applications that were previously impossible, which expanded the total market far beyond what anyone anticipated.
We're watching the same dynamic unfold in software development.
As AI reduces the cost of creating software, demand isn't contracting—it's exploding. The custom software development market, valued at $43 billion in 2024, is projected to reach $146 billion by 2030. SME adoption is growing at a rate of 21% annually, faster than any other segment. This isn't despite AI efficiency gains—it's because of them.
When development becomes faster and more predictable, projects that were previously unviable suddenly become economically viable. Use cases that couldn't justify the cost at $200/hour become compelling at $80/hour. Companies that could only afford one system now build three. The efficiency gains don't reduce total spending—they unlock latent demand that was always there, waiting for the economics to work.
In 2024, Klarna made a decision that sent tremors through the enterprise software industry. The fintech giant eliminated over 1,200 SaaS tools, replacing Salesforce, Workday, and countless other vendors with custom-built, AI-powered internal systems.
The immediate savings were substantial—over $40 million annually. But that misses the strategic point.
What Klarna recognized is that AI has collapsed the cost of building what you used to have to buy. Their AI assistant, Kiki, now handles 2,000 internal inquiries daily. Their customer service AI performs the equivalent work of 700 full-time employees. And all of this runs on infrastructure they control, generating proprietary data they own.
CEO Sebastian Siemiatkowski put it directly: they're building an AI-native enterprise from the ground up. Not by layering AI onto existing systems, but by reconstructing their entire tech stack around what AI makes newly possible.
This isn't just a cost play. It's a competitive moat. When your systems are custom-built on data you generate, you're not just running better—you're accumulating advantages that compound over time.
The traditional software market created a barbell distribution. Large enterprises could afford custom development. Small businesses used cheap, commoditized tools. Mid-size companies—too sophisticated for basic solutions, too constrained for enterprise custom development—were stuck in an awkward middle ground.
AI-assisted engineering breaks this pattern.
When development costs drop by 40-60%, the threshold for viable custom projects moves dramatically. A $500K project that only made sense for companies with $100M+ in revenue suddenly pencils out at $50M. A system that required 18 months and 10 developers can now be delivered in 6 months by a team of four.
Mid-size companies have always known what they needed. They just couldn't afford it. That constraint is dissolving.
The traditional model of selling software development—counting hours and bodies—is becoming obsolete. When AI compresses a 100-hour task into 20 hours, billing by the hour becomes a structural disadvantage for everyone involved.
The shift is toward productized, outcome-based delivery. Instead of circular conversations about team size and hourly rates, the discussion becomes: What business problem are we solving? What's the measurable impact? What's the value of that outcome to your organization?
This transformation benefits everyone. Clients get predictable costs tied to business results. Development partners can scale through intellectual capital rather than headcount. And the alignment shifts from managing inputs to delivering value.
The companies that will thrive in this environment are those that can package AI-accelerated development into reliable, repeatable solutions. Not hourly rate suppliers, but outcome delivery systems.
We're watching the old web stack fracture. For decades, developers built tribal identities around languages and frameworks. Ruby vs. Python. React vs. Vue. These loyalties made sense when mastering a language took years and switching costs were high.
AI-assisted code generation is rendering these distinctions increasingly irrelevant.
When an LLM can generate competent code in any mainstream language, the question isn't "what can our team write?" It's "what runtime best serves our requirements?" The barrier shifts from language mastery to architectural judgment.
What matters now: how reliable is the runtime? How inspectable is the system under load? How well does it scale? How efficiently can context be transformed into working systems?
We've placed our bet on Elixir and the BEAM ecosystem—not because of language preference, but because of runtime reliability. The reasons are pragmatic: OTP delivers proven fault tolerance at scale, the actor model provides natural isolation and self-healing, and the runtime offers observability that other systems can't match.
Elixir was designed decades ahead of this curve. Its strengths—stability, observability, concurrency, introspectability—match exactly what AI-generated systems need: a runtime that self-heals, isolates failures, supervises processes, and keeps systems alive under pressure.
In a world of increasingly disposable software, where AI makes creation cheap, what becomes valuable is accountability. Systems that can be trusted, inspected, and relied upon. That's not about language preference—it's about engineering discipline.
As AI commoditizes software creation, what remains scarce? Proprietary data. Context. The accumulated knowledge of how your specific business operates.
Companies following Klarna's lead understand this. When you run custom software in your environment, you don't just get functionality—you generate data. That data trains models that improve your operations. Those improvements generate more data. The loop compounds.
SaaS vendors, by contrast, aggregate your data with everyone else's. Your competitive insights dilute into their general training. Your operational patterns become their product improvements for your competitors.
Custom software built on AI-assisted development enables something that was previously too expensive for most companies: proprietary intelligence infrastructure. Not AI as a feature, but AI as a continuously improving asset that becomes harder for competitors to replicate over time.
The line has moved. Here's how to think about it now:
Buy when:
Build when:
AI-assisted engineering shifts this threshold dramatically. The projects that were firmly in "buy" territory 24 months ago are increasingly viable custom builds. For mid-size companies, this represents the most significant strategic opportunity in a generation.
The expectations have shifted. Clients aren't just buying code—they're buying solutions to business problems with measurable outcomes.
Two objections have kept mid-size companies away from custom development for years.
The combination of lower costs and higher predictability removes the two biggest barriers that kept mid-size companies locked into off-the-shelf solutions they've long outgrown.
AI is disrupting everything, especially how software is built. The old web stack is under threat. Framework loyalties are becoming irrelevant. Only the tools that help teams solve tomorrow's problems—faster, more reliably, more economically—will matter.
For mid-size companies that have long been priced out of custom development, this is the moment. The cost curves have shifted. The predictability has improved. The partnership models have evolved.
Companies that recognize this shift early will build proprietary systems and data moats while their competitors are still negotiating SaaS contracts. Those who wait will find themselves trying to catch up with organizations that have been compounding advantages for years.
It's: What is the software you can uniquely own that no one else can buy?
And for the first time in history, that question has an affordable answer.
At Appunite, we're building the systems, teams, and partnerships to help you find it.
The companies and agencies that thrive in the next decade won't be the ones with the cheapest developers or the fanciest AI tools. They'll be the ones who understand that in a world where anyone can generate code, the ability to generate reliable, context-aware systems becomes the ultimate competitive moat.
