The Future of Software Engineering in an AI-Native World
Ai generated image

The Future of Software Engineering in an AI-Native World

I recently watched an interview with Boris Cherny on Lenny’s Podcast, and I wanted to share my take on it because it put words to something I have been seeing play out in real time.

If you don’t know Boris, he is the Creator and Head of Claude Code at Anthropic. The conversation focused on AI and coding, but what really stood out to me is that it was not just about tools. It was about a fundamental shift in how software gets built.

As VP of Engineering at ConnectWise, I am right in the middle of driving hyper-automation and AI adoption across teams. This discussion did not feel theoretical. It felt like a description of what is already happening.

The Core Thesis: Coding Is Being Solved

Boris said something that stuck with me:

We are approaching a world where coding itself is “solved” by AI, and that fundamentally changes software engineering and knowledge work.

That is a bold statement, but it tracks.

We are not fully there yet, but we are clearly moving in that direction. AI is already capable of generating large portions of working code. More importantly, developers are starting to trust it enough to make it part of their default workflow.

This is not a tooling shift.

This is a paradigm shift.

What’s Happening Right Now

What makes this moment different is how fast it is happening.

This is not like previous waves of developer tooling where adoption took years. AI-assisted development is scaling rapidly because it removes friction immediately. Once engineers see that it works, they do not go back.

We are seeing:

  • AI generating meaningful portions of code
  • Developers spending less time typing and more time directing
  • Teams experimenting faster than ever before

We are moving from AI-assisted development to AI-first development.

The Printing Press Moment

Boris used an analogy that really clicked for me.

The printing press did not just make writing faster. It democratized communication.

AI is doing the same for code.

We are no longer limited by how fast someone can type or how many engineers we can hire. Code is becoming a lever, not a bottleneck.

And when that happens, the value shifts.

The Shift in Engineering Value

If AI can generate code, then writing code is no longer the highest value activity.

The value moves up the stack.

From:

  • Writing code
  • Debugging implementation details

To:

  • Defining problems clearly
  • Designing systems and architecture
  • Setting constraints and guardrails
  • Orchestrating how AI systems work together

I am already seeing this across our teams at ConnectWise.

The engineers who are thriving are not the ones writing the most code. They are the ones who think in systems and know how to guide AI toward the right outcome.

The New Role of the Engineer

The role of the engineer is evolving into something closer to:

  • Problem Architect
  • AI Orchestrator
  • System Designer
  • Quality Gatekeeper

We are moving from builders to directors.

Instead of writing every line, we are defining intent, reviewing outputs, iterating quickly, and ensuring quality and safety.

It is a shift from execution to amplification.

The Winning Operating Model

Beyond small teams, Boris outlined what I would call the winning operating model for this new world.

  • AI-first development: AI is the default starting point.
  • Rapid experimentation: Speed matters more than perfection. We iterate fast and learn fast.
  • Low-friction environments: Teams need access to tools, fewer approvals, and minimal blockers.
  • Small, high-leverage teams: AI amplifies output, so fewer engineers can do more.

We are leaning into this at ConnectWise.

We have identified key teams that are running pilot programs to demonstrate how AI can safely accelerate development. The goal is not just speed. It is proving that this model works at scale and responsibly.

Emerging Risks

With all of this acceleration, there are real risks.

  • Over-reliance on AI-generated output
  • Variability in code quality
  • Reduced visibility into implementation
  • Security and compliance concerns

As we push forward, we are working closely with our InfoSec and Compliance teams to ensure strong validation, human oversight, and secure usage patterns.

Speed without control is not a strategy.

What Comes Next

If coding becomes easy, the next frontier is clear.

  • Multi-agent systems
  • Autonomous workflows
  • Human and AI collaboration models
  • End-to-end orchestration of intelligent systems

We are not just building applications anymore.

We are building systems of agents.

Strategic Implications for Leadership

If you are leading engineering teams today, here are five steps I would strongly recommend:

1. Redefine engineering productivity: Measure outcomes, cycle time, and AI-leveraged output.

2. Invest in an internal AI agent platform: Enable agent creation, tool integration, and workflow orchestration.

3. Stand up small agent teams: Focus on high-impact use cases and prove value quickly.

4. Prioritize governance early: Build validation, security, and compliance into the foundation.

5. Upskill your organization: Train teams on AI interaction, system thinking, and evaluation of outputs.

My Takeaways

If I had to simplify it:

  • Engineering is shifting up the stack
  • AI-first workflows will dominate
  • Small, fast teams will win
  • Agent-based systems are the future
  • Leadership must focus on multiplying output

Final Thought

We are at one of those rare moments where the ground is shifting beneath us.

This is not about coding faster.

It is about rethinking how software gets built entirely.

The leaders who recognize that early and act on it will have a massive advantage.

You can also read this article on Medium: https://medium.com/@prock13/the-future-of-software-engineering-in-an-ai-native-world-709e14696110


The shift from "engineer who writes code" to "engineer who designs systems" has been talked about for years — the difference now is that the people who don’t make that shift are visibly slower, not just theoretically slower.

To view or add a comment, sign in

More articles by Pablo Arocha

Others also viewed

Explore content categories