Case Study – Hypercode, the future of software

Hypercode – The kiwi code generator supercharging innovation

New Zealanders are well known for their ingenuity, often achieving greater bang for buck than comparable businesses in larger markets.

This innovative culture is evident across most sectors, however obtaining commensurate resourcing is a common barrier to commercialisation. There are many success stories, but there could be a lot more.

Hypercode is the new code generator from Applicable that’s doing it’s part to move the needle by slashing the time and cost of developing custom software.

The problem

Where people’s innovative ideas require software development to bring them to life, the biggest problem is very often the cost.

Since its inception in 2008 the team at Applicable have repeatedly been approached by people with ground breaking ideas; things to revolutionize their customer’s experience, radically improve efficiencies or create an exciting new software product in their vertical. But so often these ideas have been stymied by the potential costs.

This is why we frequently say, “our biggest competitor is people doing nothing”. Which is unfortunate, because doing nothing takes our economy nowhere.

Market research phase

The team at Applicable have long been working to reduce the barriers to software development, and as part of that spent much of 2022 researching no-code and low-code solutions, including code generators.

Luna Tang, Alex Chang and others tried out many of the available products, from high-end to low-end, developer focused to DIY, open source to proprietary. In the end the team were surprised to discover that everything they evaluated was either too limited, locked you into a vendor, used a poor choice of tech, or came with high license fees. Some tools were appropriate for simple DIY solutions, however as software developers the Applicable team didn’t want to be stuck with the many limitations of those.

Proof of concept

By the start of 2023 one of the team, Nick Sheffield, believed he’d conceptualized an elegant way for Applicable to build it’s own code generator. During one of the wettest summers on record, that included two rounds of flooding in Auckland and other parts of the country, Nick created an initial proof of concept.

There were a number of important ideas embodied from the start. One was to focus only on the server-side part of applications, as this tends to be data centric, predictable and boring to code. The front-end of applications, the browser or mobile app parts, are where the user experience is created, thus Applicable believes in sticking with custom for those.

Another key to success was to avoid reinventing the wheel; Hypercode was designed from the outset to leverage, and work in with, all the great modern tools developers like to use anyway. Not only did this reduce the scope of work down to just bringing everything together and adding what makes Hypercode special, but it also means it fits like a glove for developers.

Build phase

Following the enthusiastic response of team members to his proof of concept, Nick then continued to build out Hypercode over the course of 2023.

Although there was some assistance from others, in particular Alex Chang building the accompanying Hypercode Admin area, the build phase largely demonstrated something counterintuitive to today’s thinking around projects. We have become obsessed with teams and collaboration, which have their good points, but history teaches us that the early phases of invention often work best when a person quietly works away at something on their own. Of course feedback is important at a certain point, as is bringing on the velocity of a larger team, but never underestimate the potential of one motivated genius given the freedom to invent.

The disruptive advent of generative AI

2023 was of course also the year that generative AI become commonplace and the team at Applicable were, right from the start of that year, testing the coding capabilities of things like ChatGPT and Github Copilot, to see whether they replaced the need for a traditional code generator. In short, they did not.

AI tools can generate sections of code, however even in that capacity their usefulness is limited by a lack of predictability. Once an application has been developed and you’re iterating, you don’t want your code to be written differently at every iteration; the process needs to be determinative and knowable.

The Applicable team did however find some great uses for AI which they integrated into Hypercode; getting AI to help suggest new data models and introducing the ability to ask plain english questions of a database and get plain english answers back.

Maturity

The first Hypercode project kicked off in November 2023 and was delivered in January 2024, on time, on budget and to client feedback saying that it achieved just what they envisaged. It was also delivered at around 50% of the cost we would have estimated prior to the use of this tool. This was the exciting moment where all the theory proved itself to be true in the real world.

At the time of this case study, in late May 2024, a further Hypercode project is galloping along and another is poised to start. Each new project is progressively larger and more sophisticated, so we expect Hypercode to continue to be matured over the rest of the year.

The future of software development

Since ChatGPT arrived with at least a degree of ability to write code and converse meaningfully with humans, our director, John Halvorsen-Jones, has been saying that at some point in the future software development agencies will probably be replaced by AI consultants that can just work through people’s requirements, create software, obtain user feedback, adjust and deploy. At that point few humans will ever need to write code. However that future could easily be 5-10 years away and, in the meantime, there is still a great need to increase the speed and efficiency of software development.

Applicable’s dream is to democratize the creation of software by bringing the cost down to levels that most businesses can afford, enabling a lot more people to bring their great ideas to life. And hey, maybe Hypercode integrated with AI can become that future tool John envisaged.

Watch this space!