No AI was harmed… or used in writing this post.

Every so often, new tools come in that change the way the industry works. As we head into 2026, we may be in the midst of a fundamental shift in Software Engineering as a craft, and what it takes to build software.

A lot of new things are discovered in the world of AI-assisted Software Engineering every week. Some recent developments around Claude Code, Codex, Antigravity and newer models like Opus 4.5 or GPT 5.2 have made me pause and think about what it means to how my team works on a day-to-day basis. 

This article is written as of the first week of January 2026. In hindsight, a lot of content here might have changed by the time you are reading this - new research, lived experiences of developers and new products might reshape our landscape in the next 18 months. These thoughts come from frequent conversation with colleagues, friends as well as family asking “What are your thoughts on the impact of AI for your line of work?”.

As an Engineering Leader, I shared a shorter version of this with my teams. In the 1:1s that followed, I received feedback that it was “inspiring”, so I thought of putting it out there for the world to see.

Detecting Noise vs Signal: We are in an evolving landscape. The adoption of new technologies leads to innovation but can also drag you down in endless loops if you don’t experiment, learn, make mistakes, re-learn and understand what each model and tool gives you. Learning by “doing” is the best way I learn; find yours. Detecting what information is noise vs what data truly matters may change what you are delivering. Clarity is the single most important factor that can decide outcomes.

Agency: You pick up a ticket, follow existing established patterns & implement just what you think is the right solution within the constraints - this may be a thing of the past with the new models that have been built through reinforced learning on codebases. “Agency” - taking ownership of a problem, coming up with solutions, analyzing tradeoffs and seeing things through to completion - is a highly desirable skill. Reflect on what you need to do to learn new non-technical skills or sharpen your existing skills. I see using AI-assisted coding as a net positive to anyone who can iterate ambiguity into smaller chunks of work to complete and are largely self-driven. 

Everyone Started at 0: When a new thing comes out, there’s a fear of missing out (FOMO). Don’t worry too much about someone being more of an “expert” than you. The key is that everyone is discovering it together, so focus on learning & experimenting. 

Tokens are $$$ too: Burning through tokens may be a good way to learn, but understand the costs that come up in how you use them. Learn how context windows work and understand that it is okay to burn through tokens; but if you find yourself burning through a lot, pause and reflect on what you may need to learn to effectively use tokens. I certainly did. Context Engineering is an emerging and fascinating area. Developers who learn how to manage limits, develop workflows against a budget for running a business will add immense value to any team. 

Follow the chatter, not the hype: On X, blogs, podcasts, YouTube, everyone is dissecting the latest AI disruptor. Startups are trying to survive in a projected multi-billion dollar AI software, infra and services market. The threat of disruption to existing incumbents is interesting. I personally don’t think a vibe-coded app for personal use can replace the data access, security, identity management, and infrastructure provided by companies providing software services. My take? Entire new categories of software will emerge, along with app marketplaces to share them in a Roblox like era of SaaS; each solving a niche problem. My current thinking is there will be a solid two years before the dust settles in this evolution around our craft. You might find out some of the same skills that made you successful before are still relevant, and you will need to adapt to new paradigms of thinking as they emerge.

Open Platforms for Humans and Robots: Platforms offer a window into the future. Ergonomic APIs that provide enough documentation to app developers (aka humans) AND robots (aka Agents) will be key to unlocking value for any business. I’m thankful that the age-old API & Integrations layer is seeing this spike in interest.

Security & Eval: Expect to review any AI-generated work to ask questions like “Is this better than what I would have written?”, “Did the agent follow safe security practices”, “Is this accurate in following my spec”, “What tests do I now need to add”, “Can this be maintained?”. Verifying your work, or better yet having another AI model run through your work will happen a lot more than you think.

Solving problems - going back to roots: I got into this career because I enjoyed solving problems with technology. From assembling my first computer to building websites,learning sound engineering and making music with it, the joy I get from being a technologist has played a big role in leading a fulfilling life. Every step of the way, I focused on solving the next problem I encountered, building software based on an idea out of thin air or installing software someone else built to make my life easier. The emergence of AI-assisted coding helps us focus on solving problems, and leaving some of the heavy lifting tasks to our AI pair-programming partner. I personally embrace this change. You really can just do things.

AI Slop Comes to Coding: A funny thing about AI chat interfaces since ChatGPT emerged is the proliferation of “AI slop” everywhere. You can see the engagement on LinkedIn, or other social media as well as traditional TV media in some countries using generative AI. Expect to see a lot more half-baked prototypes, mobile apps or entire companies built that make you cringe. It is becoming hard to figure out which OSS contribution was human vs which one was generated by AI. Sometimes I wonder though “Will it matter?” Think about economies of scale. Mass-production has evolved over time; you get varying levels of quality based on price and the amount of care the manufacturer has put in. 

You are the Product: As an Engineering Leader having conversations about AI replacing jobs, I know this is a sensitive topic. I do not have bold predictions to make. My advice has been consistent: invest in yourself as you would invest in a product or service. Being able to write about your work, having high agency to identify problems to solve, owning outcomes and building connections is more important now. The age of AI-assisted coding requires developers to adapt to being able to market their skills. As a manager, I’ve often told my team that I am their coach as well as their agent, running PR advocacy for them in rooms they are not present. The opportunities that these new tools provide are endless when it comes to showing your work.

I’m hoping some of these seem useful to you. If you’ve come this far, thank you for your patience and I'd love to hear your take. The future excites me and adventuture is out there!

What does 2026 and beyond holds for the craft that we all love?