Software, AI & Saas Business Model

How will we develop apps in the future?

The question is not so much whether or not AI will be involved in the process of software development in the future – because we know it will be to some extent. The question is, what will be the implication of AI to both software development and the Saas business model?

For context, only 4% of Saas companies achieve $1 Million in ARR (Annual Recurring Revenue) and only 0.4% achieve $10 Million in ARR.

The odds are not in the favor of Saas’ers and their investors, but the rewards for those who hit it out of the park are worth it, as we see with a lot of publicly-traded, billion-dollar companies.

Given how incentives shape behaviors, the inception of AI and its ability to either:

  • a) aid non-coders in building software, or
  • b) completely write sections of code

We can assume that there will certainly be people who try to ‘crack the Saas code’ with AI help, and thereby will potentially save money in the process or reduce the need to raise capital.

The Saas business model is likely to be transformed significantly in the next 5 years.

To illustrate the concept, we look at the formation of Canva, started by a creative couple in Australia in 2007. Canva went from $0 to $40 Billion in under a decade.


Software, AI, & Saas Business Model


Senior Software Developers and Computer Scientists are the créme de la créme of the current job market, both in terms of brainpower and the salaries they command.

Hiring Developers rarely comes cheap, and outsourcing software development comes with a host of risks; therefore, having a simple “application” (some form of code-generating AI) will certainly tempt many into ‘trying’ it on their own.

Ethical concerns aside (for now):

  • what possible advantages will there be to launching a Saas via AI-generated code?
  • what are the disadvantages?
  • will the Saas(Software-as-a-Service) business model change – structurally – due to AI-generated code, or will the market continue on as it has for the last decade+?

Let’s dive in.

Software Design – AI vs. Human

The first question about software is always what to design?

From the perspective of the market, this is always the hardest question to answer. It is a question that strikes at the heart of entrepreneurship.

There is an art to determining what software to design, as there are a million permutations and considerations. It requires abstract, multi-layered thinking, and it isn’t always coders who are the best software designers. Steve Jobs comes to mind.

Consider the creation of Canva by female entrepreneur Melanie Perkins and her now husband, Cliff Obrecht. There were three layers to the decision-making process to launch Canva in the firstplace:

1) The Need

What was the need for Canva and was it obvious?

Before Canva, amateurs had to stitch together designs in Microsoft Word or pay through the nose for confusing professional tools.


It is clear now that there was a massive need for Canva, but Perkins was unlikely to be the only one who saw the problem. Hindsight is twenty-twenty, but at the time there were probably many doubts about whether or not capitalizing on that need was worthy of the risk in capital.

Her biggest validation came from the running of their original startup, which we expand upon on below, but by no means was it a sure thing.


2) The Investment

Perkins needed investment from American VCs (she lived in Australia) to be able to design the Canva software in the first place. In her first run, she had over 100 meetings with zero bites.

Fusion Books Pitch Deck
Medium – Original Fusion Books VC Pitch Deck

Getting a meeting—much less funding—was proving tough. Perkins heard “no” from more than 100 investors.


The original idea was called “Fusion Books.” The team hired a local development agency in 2007 to build the prototype in Flash.

Perkin’s mom was a school teacher, which allowed her to see how much effort and frustration went into yearbooks. The target market for Fusion Books would be schools!

The business reached hundreds of schools, but it was not a Saas business model, and there was not enough profit to invest into the Canva vision:

The software would be free & schools would pay for printed yearbook copies.


After starting the fundraising process in 2010, it wasn’t until 2013 that their initial seed round was raised ($3 Million), and all that was after a series of fortunate coincidences around a kitesurfing workshop.

During the 3 years waiting for cheques, no real progress was made on the Canva codebase. Upon receiving the capital, they began to rewrite the code for what would become Canva in HTML5.

3) The Design

Even with a market need validated and investment capital in the bank, turning Canva into a $40 Billion dollar Saas company (at its peak in 2021) was all about design.

Source: LinkedIn

Whether it was because it was a female-driven design (more than half of Canva users are female) – or other factors – is open to debate.

After receiving some favorable reviews on tech blogs during their launch period in August 2013, user growth went vertical from there:

The trickle of sign-ups grew to 50,000 users in the first month


From the beginning, Canva was a design problem as opposed to a coding problem:

People would have to spend an entire semester learning where the buttons were and that seemed completely ridiculous. I thought that in the future, it was all going to be online and collaborative and much, much simpler than these really hard tools.

CNBC Interview

Canva’s design, price point, and market timing, turned it into one of those coveted one-in-a-million companies that everyone dreams of investing in.

Could Canva have been designed by AI?

That seems very unlikely, if not downright impossible.

The long journey of 5+ years from Fusion Books to Canva shaped the company into the behemoth it is today. The abstract levels of thinking, experience in the market, small details, etc. How would a machine achieve this?

What stands out in the story, however, is the long wait time between Fusion Books and Canva, and the need to wait for a seed round before starting the re-coding.

Software Development – AI vs. Human

Given the circumstance around Fusion Books:

  • a thriving business built on a rocky software platform (Flash)
  • a vision to solve a much bigger problem
  • being geographically located in Australia, far away from Silicon Valley

Would AI Coding Software have been tempting for the would-be founders of Canva to dabble in?

The answer is probably yes.

But would an AI Coding App be able to code an application like Canva and ensure its scalability?

There will be a lot of ‘depends’ on that question, but the most probable answer is, no, at least today.

Ultimately, finding a technical CTO was required to get the investment.

Coding a Prototype

The original Fusion Books ‘prototype’ was designed by an Australian software design agency for approximately $50,000 AUD. The $50K was raised as a loan from friends and family.

Medium – original Fusion Books software

$50K is no small amount of money to throw around for a software prototype for most people. This creates a strong incentive for would-be, non-technical software entrepreneurs to scrape together knowledge from the various ‘No Code‘ courses and start playing around with various tools, including AI Coding Apps.

Practically speaking, underfunded, non-technical founders will have a strong incentive to dabble in AI software development for the creation of prototypes.

Coding a Scalable App

The major roadblock to get Canva funded and running was finding a Technical co-founder. Software investor Bill Tai – the first investor in Canva – wouldn’t invest until one was found. That didn’t happen until 2012:

Cameron Adams joined the team in 2012. Lars had introduced him earlier, but he had been focused on building Fluent Mail. Bill agreed to invest ($25k to start) and help the team raise more capital for Canva’s MVP.


In the future, will investors invests serious capital into AI-created software?

While theoretically possible, the mountain of possible problems that can emerge in software development mean that someone (a human) will need to be there to at least audit, validate, and monitor the code.

The argument for AI-driven software development cost, theoretically, is to lower capital costs towards developing new software/apps. The reality, however, is about confidence in who is behind the software as much as the underlying code itself.

For building a scalable software app, we need humans.

Software Testing – AI vs. Human

As we have seen above, for all intents and purposes, Canva was launched in 2007.

They had 116 schools signed up in the first year, and scaled-up to several hundred more schools in the subsequent years.

That means in the five years before Canva was officially launched, the founders had probably thousands of hours of experience in testing, finding out what worked, and what didn’t work.

Humans have an innate ‘feel’ for what should or shouldn’t be in an application, both functionality and design wise. A lot of times, the tendency is to overcomplicate the design or add too many features.

Machines have no feel and are all about data. They can index what works and what doesn’t work, but are ‘soulless.’

Both feel and data are important to building winning software applications, so there may be room in the future for a hybrid approach to testing.

Trusting the Application and Its Intent

No matter how the application is coded, to gain a scalable user base and reach the vaunted 4% – or like Canva, the 0.4% – users need to trust both the application and the intent behind the application.

If a bot is designing a piece of software and people don’t trust its intentions, it doesn’t matter how flawless the code is. On the flip side, if the intentions are pure and there is market demand, a buggy app will doom the business.

Other research on AI shows distrust humans have in content written by AIs. Will code written by AIs be met with similar distrust and ultimately contempt?

It is a risk that those who enter the market must take, time will tell and disclosures for code will be tricky.

Learn to Code? Yes

As we can see above, there is a certain inevitability to ‘AI coding’ in the future.

But the need for real people to code will remain constant. If anything, AI Coding Software will probably bring more people into the coding market.

Yet only those with real experience, expertise, and products shipped will be trusted to launch major new softwares and Saas business’s into the market. At least where major capital is concerned.

There will likely be some experiments by non-technical founders using AI, No Code (already happening), and other tools. But where capital is concerned, real expertise is necessary.

The Saas Business Model

Like everything linked to AI, there will be a lot of controversy.

Given the long odds of building a scalable Saas business and the increasing economic tensions in the market, any new methods to build Saas businesses for cheaper will obviously be tested.

The Canva example illustrates how a lot of the really successful Saas businesses originate. Years of trial and error, frustration, and ultimately some luck. Few solid Saas business’s are overnight successes.

If, in the end, AI Coding Software lowers barriers to entry and ultimately makes the market for software development cheaper to access for non-technical founders, then some new Saas products may end up being more affordable to launch.

More affordable products translates to more competition in the marketplace, which will put pressure on margins. Interestingly, Canva illustrates what can be expected in markets when a cheaper and more accessible Saas alternative emerges.

Therefore, AI and the corresponding coding applications enabled by it may help ‘unbundle’ the Saas market and make it more competitive. Ethical concerns and trustability will be paramount, but what will likely happen is that hybrid solutions will emerge.

The Saas business model is likely to be transformed significantly in the next 5 years.

More AI Content

Sales, AI & Business Writing

Creatives, AI & Visual Media

Search, AI & Product Discovery