It’s easy to see why anxiety around AI is growing—especially in engineering circles. If you’re a software engineer, you’ve probably seen the headlines: AI is coming for your job.

That fear, while understandable, does not reflect how these systems actually work today, or where they’re realistically heading in the near term.

Despite the noise, agentic AI is still confined to deterministic systems. It can write, refactor, and validate code. It can reason through patterns. But the moment ambiguity enters the equation—where human priorities shift, where trade-offs aren’t binary, where empathy and interpretation are required—it falls short.

Real engineering isn’t just deterministic. And building products isn’t just about code. It’s about context—strategic, human, and situational—and right now, AI doesn’t carry that.

Agentic AI as it exists today

Today’s agentic AI is highly capable within a narrow frame. It excels in environments where expectations are clearly defined, rules are prescriptive, and goals are structurally consistent. If you need code analyzed, a test written, or a bug flagged based on past patterns, it delivers.

These systems operate like trains on fixed tracks: fast, efficient, and capable of navigating anywhere tracks are laid. But when the business shifts direction—or strategic bias changes—AI agents stay on course, unaware the destination has moved.

Sure, they will produce output, but their contribution will either be sideways or negative instead of progressing forward, in sync with where the company is going.

Strategy is not a closed system

Engineering doesn’t happen in isolation. It happens in response to business strategy—which informs product direction, which informs technical priorities. Each of these layers introduces new bias, interpretation, and human decision-making.

And those decisions aren’t fixed. They shift with urgency, with leadership, with customer needs. A strategy change doesn’t cascade neatly through the organization as a deterministic update. It arrives in fragments: a leadership announcement here, a customer call there, a hallway chat, a Slack thread, a one-on-one meeting.

That’s where interpretation happens. One engineer might ask, “What does this shift mean for what’s on my plate this week?” Faced with the same question, another engineer might answer it differently. That kind of local, interpretive decision-making is how strategic bias actually takes effect across teams. And it doesn’t scale cleanly.

Agentic AI simply isn’t built to work that way—at least not yet.

Strategic context is missing from agentic systems

To evolve, agentic AI needs to operate on more than static logic. It must carry context—strategic, directional, and evolving.

That means not just answering what a function does, but asking whether it still matters. Whether the initiative it belongs to is still prioritized. Whether this piece of work reflects the latest shift in customer urgency or product positioning.

Today’s AI tools are disconnected from that layer. They don’t ingest the cues that product managers, designers, or tech leads act on instinctively. They don’t absorb the cascade of a realignment and respond accordingly.

Until they do, these systems will remain deterministic helpers—not true collaborators.

What we should be building toward

To be clear, the opportunity isn’t to replace humans. It’s to elevate them—not just by offloading execution, but by respecting the human perspective at the core of every product that matters.

The more agentic AI can handle the undifferentiated heavy lifting—the tedious, mechanical, repeatable parts of engineering—the more space we create for humans to focus on what matters: building beautiful things, solving hard problems, and designing for impact.

Let AI scaffold, surface, validate. Let humans interpret, steer, and create—with intent, urgency, and care.

To get there, we need agentic systems that don’t just operate in code bases, but operate in context. We need systems that understand not just what’s written, but what’s changing. We need systems that update their perspective as priorities evolve.

Because the goal isn’t just automation. It’s better alignment, better use of our time, and better outcomes for the people who use what we build.

And that means building tools that don’t just read code, but that understand what we’re building, who it’s for, what’s at stake, and why it matters.

—

New Tech Forum provides a venue for technology leaders—including vendors and other outside contributors—to explore and discuss emerging enterprise technology in unprecedented depth and breadth. The selection is subjective, based on our pick of the technologies we believe to be important and of greatest interest to InfoWorld readers. InfoWorld does not accept marketing collateral for publication and reserves the right to edit all contributed content. Send all inquiries to doug_dineley@foundryco.com.