From the top of the hype curve to everyday work: how AI will change software developers’ work

From a perspective of a frontend focused developer, the coming years will bring a shift from websites/apps focused development to API focused developments – not talking just about REST APIs.

Operating systems such as Windows, macOS, Android and iOS plus the newcomers that are exclusively built for that era (for example, Rabbit R1) will introduce a way to "plug in" your application directly into a unified User Interface, where users can interact with it in their preferred way, be it text, gestures or voice.

Users will increasingly expect that they could just directly ask what they need directly, instead of finding an app/website and spending time on learning how its UI works. Instead they would prefer to just say:

• "Order groceries from Alepa. Order the same stuff that I usually get."

• "Book a table in restaurant X on Friday"

• "Give me an overview of equipment that needs maintenance. Order maintenance only for equipment that we can't continue our work without. Schedule gradual maintenance plan for the rest of the equipment".

I know, we've been promised this with existing assistants, but that technology could never deliver such a feat. This time around we will have the right tools.

Business, design, security and software development experts getting together and essentially describing (in words, images, even sounds) in every detail how the software would function, how it should work together with existing systems, how it would integrate with the unified UIs of different OSes. This will probably bring a new kind of fragmentation and experts that are focused on one OS in particular.

All these fields will overlap more and more over time driving demand for such diverse experts (software developers becoming more of a business developers and business developers becoming more of a software developers). For example, check out v0, where a designer/developer can generate a slick UI and readymade code with just a few sentences.

Traditional projects are not going away anytime soon, but there will be less and less of those on the market, with fierce competition and ever falling price points. There will be more offers to do the same thing, but cheaper and quicker due to adoption of the AI based development tools.

We should expect the next wave of AI based development tools to bring a step improvement in how code is being written. Next wave of such tools, take for example the controversial Devin AI, would bring not just high quality code generation, but a complete set of AI tools to take over mundane tasks from a developer: setup, code, security, tests and deployment with a developer role being more like a pilot of a modern airplane. The plane is capable of being on autopilot for pretty much the whole journey from gate to gate and the pilot will only take control when needed.

These tools won’t be pick-and-mix for developers. It is likely that customers will be providing these tools to consultants to keep control over what's being used. Consultants will be expected to know how to use the most popular ones (think of Jira and Figma).

Backend systems will gradually shift from providing a layer between a front end and a database to analytics driven data (and eventually UI) generators. Where, based on a set of predefined business rules, real time and historic data, it'll be delivering a non-determine set of data and UIs to fulfill a request that is sent either from a traditional app/website or the new unified UI on modern Operating Systems. So it is delivering exactly what is needed in the moment – and nothing more.

Since these systems will contain less and less hardcoded logic, security will become even more important than it already is. New kinds of security experts will be required to prevent adversarial attacks, including those that are designed to slowly happen over months and even years, only to be triggered at the right time.

As mentioned above, the user's Operating System will offer a unified User Interface where other developers can plug in their services. While still offering touch and pointing device support, new UIs that incorporate voice and new kind of gesture would gain importance, so that designers and developers will not just think of how it will look and how it will respond to the user's touch, but also how it will respond to their text or voice based input..

Content (text, images, videos) generation will be done on demand and will become part of designer/developer/business workflow. Same as with the AI-based software tools, eventually it will be expected that experts are familiar with how content generation works and be able to effectively integrate that into their work.

I expect the majority of companies to keep relying on readymade cloud services (OpenAI, Google Vertex, Amazon Azure, etc). But for certain cases, e.g. where instant response is paramount, companies will opt out to build their own processing capacity. That’ll mean a lot of GPUs/TPUs here. This will also require outside expertise in designing, building, running and maintaining these systems both from hardware and software perspective.

The less connected you are with the everyday of software development, the more you’d think AI would reduce demand for developers. The productivity will rise, that’s a given. But the demand won’t fall. It will change.

As the demand for developers is likely to increase, the nature of the demand will be shifting. While AI can increase productivity by automating some coding tasks, the need for developers to oversee, integrate, and innovate with AI technologies will grow. Developers will need to adapt to a role that involves more oversight of AI tools, integration of AI into existing systems, and continuous learning to keep up with AI advancements.

Knowing what I know now, I would definitely tell my younger self to specialize in software development. The development of AI will make a lot of jobs obsolete, but not this one. We are going to be just fine.