Let Go, Let Agents: A New Shape of the Workday
A few days ago I read “Let Go, Let Claude” by Rob Conery and it genuinely stopped me in my tracks.
Not because it was futuristic.
Because it was practical.
He described something subtle but profound: the shift from doing the work yourself to designing, delegating, and refining work done by agents. I had been tinkering with markdown-driven workflows for business tasks. Specs in plain text. Structured thinking. Clean artifacts.
But his framing crystallized something I had been circling without naming.
This is not about prompting better. This is about operating differently.
From Individual Contributor to Agent Orchestrator
Here is what my new workday looked like yesterday.
Instead of diving straight into execution, I started with planning. Almost like a personal daily standup. But instead of thinking about what I would do, I thought about what my team of agents would do.
I identified four discrete tasks.
For each task, I created a spec file in markdown:
Clear objective
Constraints
Inputs
Outputs
Acceptance criteria
Planning complete.
Then I opened four separate sessions using OpenCode and Claude Code. One agent per session. Each assigned its own spec. Each operating independently.
Then I stepped away.
Coffee.
Came back. Reviewed outputs. Refined instructions. Clarified constraints. Iterated.
Met with humans (mostly, I see you note taking agents).
Lunch.
Checked on agents.
More meetings.
Final polish on outputs.
End of day.
And I had facilitated roughly four times the volume of work I could have personally executed in a traditional workflow.
Not theory. Yesterday.
All of it running locally and freely via OpenCode, Ollama, and gpt-oss:20b.
The Cognitive Shift Is Real
What surprised me most was not productivity.
It was how my brain felt.
Soggy.
The closest analogy I have is from my Black Belt days. After a full day buried in Excel, analyzing data, forming conclusions, then translating those conclusions into business narratives. That specific cognitive fatigue. High bandwidth. Context switching. Abstraction layers stacked on abstraction layers.
This felt similar.
Because the work is different.
You are not typing faster. You are thinking structurally.
You are designing task systems. You are specifying intent precisely. You are evaluating outputs critically. You are iterating at a meta-level.
It is closer to managing high-performing analysts than it is to writing code or drafting documents.
And some people will adapt quickly. Some will not.
A New Shape of the Workday
If this model continues to mature, the archetypal workday changes.
Morning: Architect the day. Define deliverables. Write specs for agents. Midday: Supervise and refine. Interleave human meetings with agent oversight. Afternoon: Integration and polish. Merge outputs. Validate coherence. Adjust tone and strategic framing. Evening: Step back having orchestrated multiples of your former throughput.
The skill is no longer raw output generation. It is orchestration, judgment, and refinement.
That requires different mental muscles.
This Is Happening Now
It is tempting to categorize this as experimental or bleeding edge.
It is not.
The tooling is already here. Free. Local. Practical.
OpenCode
Ollama
gpt-oss:20b
The barrier is not cost. It is mindset.
Letting go of the instinct to personally execute everything. Trusting systems. Writing better specifications. Thinking modularly.
In many ways, it mirrors how software engineering matured. From lone coders to distributed systems thinkers. From scripts to architectures.
Now that shift is happening at the individual knowledge worker level.
What This Means for Executives
For leaders, this is not about which model to buy.
It is about whether your organization is preparing people to:
Decompose work into modular tasks
Write clear specifications
Validate AI output rigorously
Think in parallel instead of serial execution
Manage cognitive load in high abstraction environments
Because this does not just change productivity.
It changes how people think.
Yesterday was my first real attempt at operating this way. It worked. It was productive. It was exhausting in a new way.
And I cannot unsee it.
Some will adapt and compound. Some will resist and stall.
But this is not a hypothetical future.
It is available right now.
And once you feel that soggy-brain shift after orchestrating multiple agents in parallel, you realize something important:
The work has already changed.