Should you learn to build AI features?

AI is no longer something happening “over there” in research labs. 

It’s embedded in the tools you use, the products you build, and increasingly, the expectations companies have for their teams. The real question isn’t whether AI is important; it’s whether you, as an engineer, should invest time in learning how to build AI features.

The answer isn’t a simple yes or no. It depends on your role, your career goals, and how you want to engage with this wave of technology.

The many layers of AI fluency

AI knowledge isn’t a binary — you don’t either “know AI” or “not know AI.” Just as not every front-end engineer needs to understand the deepest quirks of React, there are different layers of involvement:

• At the core are the PhDs and researchers building models from scratch.
• One level out are applied AI engineers who are focused on scaling, deploying, and integrating those models into systems.
• Further out are product engineers who call APIs, combine frameworks, and ship real features.
• On the edges are front-end or UI engineers, who don’t write AI code directly but still need to understand what’s possible, how features behave, and how to design around them.

The reality is that most engineers will fall somewhere in the middle. You don’t need to be a researcher to stay relevant, but you do need enough fluency to work alongside AI systems and teammates who build with them.

What the market is signaling

Even though technical interviews haven’t yet shifted wholesale to “AI-first” evaluation, the signals are clear:

• Hiring conversations already include questions about how you’re staying current with technology — and AI is always part of that.
• Startups are under pressure to ship “with AI” because funding often depends on it. That means they need engineers who can build features, not just talk about them.
• Big tech companies are hoarding talent, sometimes without clear roles defined, simply to stay ahead in the AI race.
• Inside organizations, executives are pushing teams to adopt AI, often without precise direction. The expectation is: experiment, figure it out, and make it work.

This isn’t just anecdotal. According to Aspen Tech Labs’ Q1 2025 Job Market Pulse report, there were 35,445 AI-related positions across the U.S. — up 25.2% year-over-year and 8.8% from the previous quarter. The median salary for AI roles hit $156,998 in the same period. That combination (more jobs and higher pay) shows the demand is both real and accelerating.

If you can bridge the gap between leadership’s top-down mandates and actual working features, you become indispensable.

The career upside

For engineers, this creates both risk and opportunity. On the opportunity side, AI-related roles are multiplying. 

There’s a rapidly growing set of roles for engineers who know how to build with AI:

• Applied AI engineers who focus on scaling models, managing traffic, and gluing systems together.
• AI ops specialists who handle deployment and reliability, much like DevOps for AI.
• Product engineers who can integrate APIs and frameworks into user-facing features.

And the payoff is real. PwC’s 2025 AI Jobs Barometer found that roles requiring AI skills carry, on average, a 56 percent wage premium over otherwise similar jobs. That doesn’t mean every engineer who sprinkles AI into their résumé will suddenly earn more, but it does indicate a market-wide recognition that the skillset has economic value.

The more subtle upside is relevance. When executives push for AI features and investors demand them, the engineers who can translate those demands into working products suddenly become indispensable.

The risk of standing still

Think back to the mobile revolution in the early 2010s. In the early 2010s, companies recognized the need for mobile apps. CEOs announced, often with little detail, “We need iOS.” Teams scrambled to figure out what that meant. Some engineers leaned in, became mobile specialists, and rode the wave to lasting careers. Others dismissed mobile as a side project.

In hindsight, mobile was never optional. Even engineers who never wrote an iOS app needed to understand what mobile meant for design, connectivity, and user behavior. They had to know that screen sizes were smaller, touch interactions were different, and network reliability couldn’t be assumed. Without at least that baseline awareness, their work would fall flat.

AI is in the same position now. Not every engineer will retrain as an AI researcher. But every engineer will need to understand how AI shapes products, how to work with it at some level, and how to collaborate with teams that are using it directly. To ignore it is to risk becoming the equivalent of the engineer still coding in notepad while everyone else moved on to IDEs.

Is it right for you?

That doesn’t mean every engineer should rush headlong into AI. Chasing hot fields just because they’re hot rarely ends well. During the mobile boom, plenty of people tried to pivot into iOS development for the salary bump or job security. Some stuck. Many didn’t, because they didn’t actually enjoy the work.

The same caution applies to AI. If you’re curious, enjoy working close to product, and want to future-proof your career, then learning these skills makes sense.

If AI doesn’t excite you at all, you still need basic fluency to stay effective in collaborative teams.

The decision isn’t whether AI affects you (it does). It’s more about how deep to go. Do you want to be a researcher, an applied engineer, a product integrator, or simply an informed collaborator? Each path is valid, but pretending AI doesn’t affect you isn’t.

Where to start

If you’ve decided you want more than surface-level fluency, the next step is structured practice. That’s exactly why we created Build AI-Powered Features at Formation.

The course is designed for generalist engineers — the people building APIs, dashboards, and user-facing features — who want to go one level deeper into how AI features are actually shipped. Over four weeks, you’ll move beyond demos and get hands-on experience with:

• Week 1: The features users now expect: semantic search that understands intent, smart summaries, and context-aware auto-complete.
• Week 2: Building production RAG systems: from simple retrieval to handling real product challenges across different domains and use cases.
• Week 3: Creating conversational AI: going deep on chatbot implementation, architecture decisions, and hands-on building.
• Week 4: Connecting AI to the world: using MCP to let your agents interact with tools, data sources, and systems — with practical use cases and implementation.

By the end, you won’t just “know about” AI — you’ll have built working features, developed the judgment to know when and how to use them, and gained the fluency companies increasingly expect in their engineers.

Next cohort starts October 13, 2025. Enroll here.