I just delivered a comprehensive strategic analysis to a client. The kind of deliverable you'd expect from a McKinsey or Bain engagement. Sixty thousand words of research, frameworks, recommendations, and supporting documentation. Charts, visualizations, a polished PDF.
I did it in three days. Between meetings. Between life.
The secret wasn't working harder. It was orchestrating differently.
I used Claude Cowork and Claude Code as thinking partners, researchers, and execution engines. My job was to direct, challenge, and refine. The AI handled the volume.
Here's what the process actually looked like.
Thinking and Conceptualization
The project started with me talking through the problem. Not typing a detailed prompt. Just explaining the client situation, the questions we needed to answer, and the constraints we were working within.
Claude helped me structure the approach. When I had a half-formed idea, I talked it through and let Claude push back, suggest angles I hadn't considered, and help me sharpen the framing. This wasn't delegation. It was collaborative thinking at speed.
Research and Challenge
Once we had a direction, I pointed Claude at the research. Find the relevant data. Surface the frameworks that apply. Challenge my assumptions with evidence.
The back and forth was fast. I'd review what came back, ask follow-up questions, redirect when something was off-track. The research that would normally take days compressed into hours.
Building a Source of Truth
One thing I've learned: without a structured place to capture findings, these projects become chaos. I had Claude create a data source as we went. Every key finding, every source, every data point got logged in a way that we could reference and build on.
This became the backbone of the deliverable. When it was time to write, we weren't hunting for where we'd seen a particular stat. It was all there.
A Local System of Record
As we worked, Claude Cowork and Claude Code were creating Markdown files in a local directory on my computer. Each task, each research thread, each line of thinking got captured as a discrete file. I wasn't managing this actively. It just accumulated as a byproduct of the work.
This mattered more than I expected. With minutes left before my presentation, I was able to go back through all of those files and structure the right approach. The result wasn't just one deliverable. It was a set of documents: talking points for the presentation, a research document with methodology, rankings, and comparisons, a source document with citations, and a data document capturing everything we'd used.
That local archive gave me insights I couldn't have surfaced without a team in the past. The AI didn't just help me produce. It helped me think at a level of depth and rigor that would have required multiple people and significantly more time.
Visualizations and Charts
Numbers need to be seen, not just read. I directed Claude Code to create the visualizations. Charts, frameworks, comparison tables. Each one exported as a PNG so it could drop cleanly into the final document.
This used to be the part that killed momentum. Stop writing. Open a different tool. Build the chart. Export it. Get back to writing. Now it's part of the same flow.
The Final Deliverable
Everything came together in a polished PDF. Sixty thousand words of analysis, structured into sections, supported by visuals, formatted for a client audience.
I didn't write sixty thousand words. I directed the creation of sixty thousand words. Big difference.
What This Means
Three days. Between meetings. Between life.
A year ago, I'm not sure this would have been possible for one person. If it was, it would have taken months. Today, one person orchestrating AI effectively can produce at a level that used to require a team and a timeline I couldn't have committed to.
The skill isn't prompting. It's orchestration. Knowing when to direct, when to challenge, when to let the AI run, and when to step in and refine. It's a different kind of work, but it's still work.
The people who figure out this mode of operating will have a structural advantage. The output capacity is just different.