Slack co-founder shares insights on the future of engineers and tech companies

Cal Henderson, co-founder and CTO of Slack, LinkedIn

How AI is changing the role of the engineer

I have been developing software since I was young. I have always enjoyed writing code, but even more so, creating something real.

About 10 years ago, a similar shift occurred in my career. I stopped writing code daily because all my time was spent managing people and organizing. It was a difficult transition. Previously, writing code defined my value to the company. It was my daily activity and main professional skill. But over time, I realized that I could bring much more value if I didn't write code myself. Accepting this was not easy.

Today, engineers working on products are experiencing something similar. If you enjoy creating software, launching products and services, that is what you should identify with. Writing code is just a means to achieve this goal. But when you stop doing what used to be your main professional skill, a sense of loss arises.

When I gave an interview about a year ago, the capabilities of AI in programming were completely different. The quality of the products created, their functionality, and the development process itself were also different.

Back then, working with AI agents looked like this: the agent wrote code, I studied it, suggested fixes, and provided feedback. I no longer needed to write code myself.

Today, I don't think the best engineers even continue to read code line by line. Agents do this much faster. Ultimately, the code itself is not so important — the result is. We need to accept that the process of software development has changed fundamentally.

In some ways, this is similar to the transition from assembly language to compiled programming languages. Back then, many feared that developers would stop writing assembly code, understanding the structure of their programs, and controlling the creation process. Instead, code began to be generated by compilers.

But it turned out that this was not such a big problem for development. Over time, everyone got used to working at the next level of abstraction. It was just that the transition itself was painful.

I think the same thing is happening today. But if you accept the changes and learn to work in a new way, productivity can increase many times over.

Why middle-level developers find it harder to adapt to AI

In a sense, middle-level developers are currently in the most difficult position. On the one hand, they have already learned a lot and are used to working in a certain way, so it is not easy for them to restructure. On the other hand, it is not yet entirely clear how to develop professional skills and build a career now.

Previously, there was a clear career path: it was clear what skills needed to be developed and what needed to be achieved to move forward. Now we do not yet understand what the career ladder for software developers will look like.

If you are already a lead engineer at the Principal Engineer level, AI tools give you a huge advantage. You know the fundamental principles of development, understand how systems are structured, and know how to work effectively. Now AI allows you to increase your productivity by 10 or even 100 times.

But engineers with average experience find it much more difficult. Probably, the situation will become clearer in the next couple of years, but for now, this is indeed a difficult period. Especially when you do not understand where the industry is moving, and suddenly find that the skills you spent most of your career developing no longer have their former value.

Why small teams will be able to build large businesses

Today, the most interesting AI startups are often small teams. A few people can achieve incredible things. For example, I know a biotech startup consisting of just one person who is engaged in finding new drugs. Of course, not every company should consist of one employee. But now it is possible to create and develop a business with much smaller teams than before.

This opens up opportunities for industries where it has historically been difficult to enter due to high capital, time, and personnel requirements. For example, in drug development and hardware technologies. Today, launching hardware startups has become significantly easier and faster.

Moreover, you no longer need to specifically look for companies that position themselves as AI startups. Businesses that actively use AI gain an advantage anyway: they achieve results faster, create higher-quality initial products, and use capital more efficiently.

The performance gap between engineers is becoming more noticeable

In software development, teams can already be significantly smaller. For example, employees of large technology companies, such as Slack, report how much more productive software engineers can now work.

It is software development that is currently at the forefront of AI implementation. A team of engineers that is 10 times smaller can quite possibly work more productively than a large team did a year ago.

The difference between teams that have fully integrated AI into their workflows and those that have not is colossal. A huge gap is observed even within the engineering departments of large technology companies.

Some engineers have completely restructured their approach to work and actively use modern AI tools. Others are skeptical about AI, refuse to use it, and do not believe that developers need to change their usual methods of work.

The difference in results and the volume of work completed between these two groups can reach a hundred times.

Software development has changed completely. If your knowledge of what is happening is based on experience from six months or even a year ago, it is already outdated. With the release of each new generation of models, their capabilities expand significantly, fundamentally changing the development process itself.

Why AI does not necessarily lead to mass layoffs

I do not think that large technology companies will have to reduce their staff. Imagine that you already have an operating company and suddenly every employee starts working twice as productively. What would you choose: fire half the team or increase productivity and sell twice as many products?

Obviously, you would choose the second option, especially since competitors are in the same situation. Therefore, we will most likely see increased requirements for performance and accelerated work rates, rather than mass layoffs.

At the same time, companies may increase the number of employees more slowly. Now it is important to make the most effective use of the potential of the existing team.

However, the impact of AI on different professional areas is not uniform. Some tasks can be automated and accelerated much more than others. Therefore, the transformation of different roles within companies will occur gradually and at different speeds.

For example, changes in the work of a large sales team will occur much more slowly than in software development.