AI / 6 min read
Why the World Needs More Software Engineers in the Age of AI
How AI Is Expanding Demand, Redefining Skills, and Transforming the Future of Work
Why the World Needs More Software Engineers in the Age of AI
How AI Is Expanding Demand, Redefining Skills, and Transforming the Future of Work
Artificial Intelligence is often framed as a threat to jobs — especially in software engineering. Headlines frequently suggest automation will replace developers, reduce hiring, and shrink opportunities. But a deeper look tells a more nuanced story.
In a recent conversation with Box CEO Aaron Levie, a very different perspective emerged: AI may not reduce the need for engineers — it could dramatically increase it.
Let’s explore why.
The Engineering Demand Paradox
At first glance, it seems logical: if AI makes engineers more productive, companies will need fewer of them. But in reality, something more complex is happening.
Recent data shows that software engineering and AI-related job postings are actually rising. While the long-term impact on employment is still uncertain, one insight stands out — productivity gains don’t necessarily reduce demand. They often expand it.
This phenomenon is sometimes described as a version of “Jevons paradox,” where increased efficiency leads to increased usage rather than less.
What does this mean in practice?
In the past, many software projects were too expensive or time-consuming to justify. Only large tech companies could afford to invest heavily in engineering.
Now, with AI tools boosting productivity by multiples, many of those previously “unviable” projects suddenly make business sense.
- Internal tools for small teams
- Automation for repetitive business processes
- Custom workflows across departments
Instead of reducing demand, AI is spreading it across the entire economy.
Software Is No Longer Just for Tech Teams
One of the biggest shifts happening today is where software is built — and who benefits from it.
Earlier, engineering efforts were mostly concentrated in IT departments. Today, every function in a company can benefit from software:
- Marketing teams automating campaigns
- Legal teams analyzing contracts
- Finance teams streamlining reporting
- HR teams improving employee workflows
AI-powered engineers are no longer just building products — they’re improving operations across the organisation.
Real-world example
Imagine a hospital system trying to cut costs while improving patient care. Instead of reducing staff, they use AI to automate back-office processes. This frees up time and resources that can be redirected to patient services.
In this scenario, engineers don’t replace people — they enable better outcomes.
The Real Challenge: Context, Not Connectivity
While much attention is given to integrating systems, the bigger challenge lies elsewhere: context.
Most organisations already have multiple tools and platforms that can technically connect with each other. But their data is often scattered, unstructured, and hard to use effectively.
Why does this matter?
AI agents depend on accurate, well-organised information. If data is spread across dozens of systems without structure, even the most advanced AI struggles to perform reliably.
- Too little context → poor decisions
- Too much irrelevant context → confusion
This creates a major bottleneck.
What’s coming next?
Experts predict a long phase of infrastructure modernisation, where companies focus on:
- Cleaning and structuring data
- Making information accessible in real time
- Designing systems specifically for AI workflows
Organisations that invest early in this foundation are already seeing significant productivity gains.
A New Kind of Engineering: Deterministic + Probabilistic
AI introduces a fundamental shift in how software is built. Engineers are no longer working with just predictable systems.
Today, they are managing two types of computation:
- Deterministic systems — predictable, rule-based logic
- Probabilistic systems — AI-driven, flexible, and context-aware
The key question
When should something be handled by traditional code, and when should it rely on AI?
- Loan processing → needs consistency → deterministic
- Customer support queries → benefit from flexibility → probabilistic
Making the right choice is critical — and surprisingly complex.
In fact, AI doesn’t simplify engineering. It makes it more technical, requiring deeper understanding and better judgment.
Where Startups Have the Advantage
AI is also reshaping competition.
Startups have a unique opportunity because they can build systems from scratch — without legacy constraints.
They can:
- Design AI-native workflows
- Avoid outdated processes
- Move faster than large organizations
This is especially true in areas with unstructured work, such as:
- Legal reviews
- Tax processes
- Risk analysis
- Contract management
In these domains, AI can deliver significantly higher output at lower cost.
The Hidden Challenge: Reinventing Workflows
One risk for established companies is trying to force AI into existing systems instead of rethinking how work should be done.
Organizations often resist change because:
- Roles and responsibilities are deeply ingrained
- Processes are tied to existing structures
- Teams are comfortable with current workflows
But AI works best when workflows are redesigned from the ground up.
This requires not just technical changes — but organizational ones.
Humans vs AI: The Context Gap
Humans have a natural advantage that AI lacks: context.
We understand:
- Our work environment
- Team dynamics
- Past decisions and experiences
- Unwritten rules and expectations
AI agents, on the other hand, start with zero context.
They need to be explicitly provided with the right information at the right time.
Think of AI agents as new employees
- Highly skilled
- Extremely fast
- But completely unfamiliar with your organization
To be effective, they must be “briefed” properly.
This is why structured context — clear instructions, organized data, and well-defined workflows — is becoming essential.
The Future: From Content to Context
One of the most powerful ideas from the discussion is a simple shift in perspective:
The future isn’t just about managing content — it’s about delivering context.
Businesses already have vast amounts of data — contracts, documents, reports — but much of it isn’t usable by AI.
The next evolution is making that information:
- Accessible
- Structured
- Actionable
And importantly, usable not just by humans — but by AI agents as well.