Article 002: What happens when work continues while you’re not there

Key Message: AI agents don’t support your work — they generate it. This isn’t evolution; it’s an inversion.

Series: A — Noticing the shift (Part 2 of 5) Publish Date: Monday 2026-01-26

Story Outline

Opening hook

For most of the history of knowledge work, progress required presence. You sat at your desk, you did the work, you left, it stopped. The next day you picked up where you left off. This coupling was so complete, so obvious, it didn’t need stating. Progress and presence were the same thing.

Something has changed.

The old rhythm (including CI/CD)

The fundamental model:

You do the work; automation supports it
Your keystrokes, your decisions, your creative output
Automation handles the mechanical: building, testing, deploying
But the generative work — the thinking, designing, coding — that’s you

CI/CD as evolution, not revolution:

CI/CD made the old rhythm more efficient, not different
You write code; the pipeline builds and tests it
You push a change; infrastructure deploys it
The ratio: ~80% human action, ~20% automated support
Automation serves your output; it doesn’t create output

The psychology of this model:

When you close your laptop, meaningful work stops
Systems churn through queues, but nothing new is created
There’s a restfulness in this — the project waits for you
You can count on finding it where you left it
Endings are natural; they happen when you stop
The boundary between working and not-working is clear

What CI/CD changed and what it didn’t:

Changed: speed of feedback, reliability, parallelism of supporting tasks
Didn’t change: who does the work, what requires presence, where progress comes from
You might wake to a failed build, but you don’t wake to new features
The work product is still yours

The new rhythm (AI agents)

Revolutionary, not evolutionary — an inversion:

Agents don’t support your work; they generate it
You describe intent; the agent produces code
You review and approve; the agent moves on to the next task
The work itself is created while you’re away

The ratio inverts:

Old model: ~80% human action, ~20% automated support
Current state: flipping toward ~80% AI-generated, ~20% human supervision
Trajectory: heading toward 99/1, then 99.9/0.1, then 99.99/0.01…
These aren’t predictions — they’re order of magnitude illustrations
The direction matters more than the precise numbers
We’re not talking about incremental efficiency; we’re talking about who does the work

What this looks like in practice:

You wake to pull requests you didn’t write
Features implemented overnight, awaiting your review
Bugs fixed while you slept
Refactoring completed across the codebase
The repository shows meaningful activity — not just builds passing, but work done
Decisions made, code committed, progress achieved — in your absence

The qualitative shift:

From operator to supervisor
Still creating, but at broader brush strokes and higher layers of abstraction
The agent handles fine-grained implementation; you shape the larger picture
From “doing the work” to “directing the work”
Presence becomes about judgment, not production
Your value shifts from keystrokes to decisions

Why this is different from all previous automation:

Previous automation: do this mechanical thing faster/reliably
AI agents: do this creative/cognitive thing that previously required me
The boundary of “what requires a human” is moving
Not just moving incrementally — moving rapidly and substantially

What this changes (implications)

Mornings change:

You don’t start fresh; you start by catching up
What happened overnight? What succeeded, what failed?
What’s waiting for a decision? What needs correction?
The first task is often triage, not creation
You arrive not at a blank slate but at an inbox of outcomes
The skill becomes: quickly assessing work you didn’t do

Closure becomes elusive:

There’s no natural stopping point when work continues without you
“Done for the day” becomes a choice you impose
You have to decide to stop, knowing the work won’t stop with you
The old rhythm had natural endings; this rhythm has none
Discipline shifts from “finishing what’s in front of you” to “deciding when to disengage”
Walking away requires intention it didn’t before

Attention fragments:

If work can progress at any time, notifications can arrive at any time
The boundary between working and not-working blurs
Your phone buzzes with a completed task at dinner
An agent finishes a feature while you’re watching a film
The awareness that something might need you never fully switches off
You’re always potentially on call for review/decision

The “low hum of ongoing-ness”:

A new psychological state
Not quite anxiety, not quite excitement
Background awareness that the system is alive
Work is happening, progress is being made, somewhere
Even when you’re not watching, things are moving
This hum didn’t exist before; now it’s constant

The relationship to your own work changes:

Some of “your” work isn’t yours anymore
You’re responsible for output you didn’t produce
Reviewing becomes more important than writing
Quality control becomes your primary contribution
The code has your name on it, but not your keystrokes

What you provide changes:

Constraints become essential — what should the agent not do?
Guardrails become essential — how do you keep it on track?
Vision becomes essential — what are we actually building, and why?
Direction becomes essential — what matters most right now?
Without these, agents generate noise, not progress
Your value shifts from execution to curation and judgment
The less you do, the more what you decide matters

The trade-off

What we gain:

More gets done; leverage is real
Parallelism at scale — many agents, many tasks
Work that took days can complete in hours
Start something, step away, return to find progress
Bottlenecks shift from “waiting for humans to do” to “waiting for humans to decide”
Individual output capacity multiplied dramatically
Cost of experimentation shifts dramatically
Prototyping moves from lo-fi sketches to fully working systems
Try an approach, see it built, evaluate, pivot if needed
Cost of being wrong drops; cost of exploring alternatives drops
Discovery through rapid iteration replaces careful upfront planning

What we lose or what becomes harder:

Rest requires intention it didn’t before
Boundaries require active maintenance
The skill of “putting work down” must be cultivated
Old natural endings replaced by deliberate ones
Harder to feel “done” — done with what?
The mental separation between work and life erodes

What changes in unexpected ways:

Learning and skill development — if you’re not doing, are you learning?
Debugging — how do you debug code you didn’t write?
Ownership and accountability — whose bug is it?
Onboarding — how do you learn a codebase agents maintain?
Career progression — what does “senior” mean when agents do the work?

What this doesn’t mean

Presence still matters — but for different things:

Focused, uninterrupted thinking still requires engagement
Creative leaps still require presence
Complex collaboration still requires synchronous attention
The work that continues without you is a particular kind of work
Judgment, taste, direction — these still require you

Not all work will shift at the same rate:

Routine implementation: shifting fast
Novel architecture: slower
Cross-functional coordination: requires humans
Anything requiring organisational context: still human
The frontier is uneven

This isn’t a loss of agency — but it is a change in what agency means:

You still direct; you just don’t execute
Your decisions still matter — arguably more
But the nature of your contribution changes fundamentally
From exclusive presence to mixed presence
From continuous doing to intermittent supervision

Close

Reframe the shift:

Progress is no longer tied to presence
For most of knowledge work history, that sentence wouldn’t have made sense
Now it’s simply true — not for supporting tasks, but for the work itself
Features written, bugs fixed, decisions made, all while you slept

What this changes (callback):

More than productivity
What your presence is for
What mornings feel like
What endings require
What rest means
The relationship between you and the work that carries your name

Where we are:

The rhythms haven’t settled
The boundaries are being drawn and redrawn
The low hum of ongoing-ness is new
We’re each finding our own way to live with it

Landing:

The work no longer waits for you
What that means — we’re still learning

Notes

Builds on article 001’s observation that the desktop isn’t always central
Explores why — work continues without presence
Key distinction: CI/CD = evolution (old rhythm, efficient); AI agents = revolution (new rhythm, inversion)
The “low hum of ongoing-ness” is the standout phrase
Numbers (80/20 → 20/80 → 1/99 → 0.1/99.9…) are illustrative of magnitude, not predictions
New themes: learning, debugging, ownership, career progression — worth exploring in later articles?

Review Feedback

Clear separation between old (CI/CD) and new (AI agents) — addressed in outline
Order of magnitude illustration (80/20 → beyond 99/1) — addressed in outline
More substantial sections with depth — expanded all sections
Longer, more deeply considered paragraphs (apply when writing)
Strengthen weak closer — options provided, needs decision
Check for “leave the office” → “close your laptop”
Check for “Deep work” unexplained jargon
Check for inevitability claims that need softening

Techcle Wiki

Explorer

Article