A couple of days ago, while discussing ways of “improving” delivery, a manager said:
“We can use AI to generate the tickets.
Then AI can create the pull requests.
Eventually we won’t need engineers for most of the workflow.”
No drama. No threat. Merely practical, efficient, logical.
Yet…
It felt wrong.
And the problem is not AI. Of course it can do all that — and more — faster and better.
The problem, at its core, is how this framing quietly redefines what “delivery” means.
Just think about it.
If work becomes tickets,
and progress becomes velocity,
then humans are bottlenecks.
And if that happens, replacing humans becomes optimization.
The Problem Isn’t AI — It’s What We Call “Value”
For years, many organizations have measured delivery like factory output.
Performance gets reduced to activity metrics:
- Tickets closed
- Story points completed
- Tasks finished
- Lines of code written
- Burndown/Burnup charts
But software engineering was never assembly-line work.
Those metrics were always approximations — attempts to measure creative, uncertain, deeply contextual work using industrial-era tools.
Still, they became “the KPIs.” The standard. The reference point.
Many managers proudly showcase:
- Precise dashboards
- Objective reports
- Perfect performance reviews
Even when those numbers say very little about real value.
AI doesn’t create this problem.
AI exposes it.
Because now:
- Machines can generate tickets
- Machines can write code
- Machines can create documentation
- Machines can produce output faster than any human team
So if output is how we measure value…
Humans lose by definition.
Understanding Context Is Valuable — Even When Automation Makes Activity Cheap
AI enables us to produce more:
- More code.
- More plans.
- More documentation.
- More tasks.
- More everything.
But quantity was never the real bottleneck.
Understanding IS.
Understanding context is the value.
- What’s the real problem?
- What are the tradeoffs?
- What are the long-term consequences?
- How does this decision ripple across the entire system?
AI accelerates execution of course.
But execution without understanding is how systems slowly drift in the wrong direction — while every metric says we’re improving.
- Dashboards are green, yei!!.
- Systems quietly become fragile, awww!!
Performance Theater Scales Beautifully
When managers focus on the wrong metrics:
- People optimize for visible activity instead of meaningful impact
- People close tickets instead of solving problems
- People deliver features instead of improving outcomes
- People add complexity because complexity looks like importance
And everyone can honestly say:
“Nothing to do here. MY work is DONE.”
Even if the system is worse.
AI supercharges this toxic dynamic.
Now performance theater can be automated:
- Plans generated without deep reasoning
- Code shipped without full understanding
- Roadmaps created without strategic clarity
Activity scales. Meaning doesn’t.
What Actually Worries Me
“Software engineering is at risk.”
Nah.
There will be changes — like there always have been.
The roles safest from automation are those creating meaning. The roles most exposed are those measuring activity.
AI can generate artifacts, but it cannot ensure they make sense inside a complex system.
Meaning requires context, and context lives in people who take the time to understand how things connect.
A Quick RACI Reality Check
You might be familiar with the RACI matrix:
Responsible — the ones who implement the work Accountable — the ones who own the outcome Consulted — the ones who provide input Informed — the ones kept updated
For years, many roles built their value around responsibility (execution).
Today, AI can handle more and more execution.
But AI cannot truly be Accountable.
Accountability means:
- Standing behind outcomes
- Understanding consequences
- Validating that solutions address real problems
- Owning tradeoffs
- Acting when things break
- Answering why
As automation expands, execution becomes cheaper.
Accountability becomes rarer — and far more valuable.
What Is the Real Risk?
The real risk isn’t engineers being replaced.
The deeper risk lies in environments that confuse activity with impact:
- Systems that reward motion over direction
- Metrics that measure noise instead of signal
- Cultures that optimize for speed without asking where they’re going
Automation doesn’t remove accountability.
It concentrates it.
When execution becomes easy, the real work becomes:
- Judgment
- Context
- Ethical reasoning
- Understanding system-wide consequences
- Understanding why
A Better Question
Instead of asking:
“How do we automate this workflow?”
We should also ask:
- Where must humans remain deeply involved?
- What kind of thinking should never be outsourced?
- Are we optimizing activity or outcomes?
- Does this make the system clearer — or just faster?
AI can run workflows – but it cannot understand why the workflow exists.
The accountability is still ours.
And it’s becoming more valuable, not less.