Errol is a product manager at a tech company. He attended one of our workshops and built his first AI workflow: weekly stakeholder updates. It saved him 20 minutes per week.
Most people would stop there. Twenty minutes saved isn't life-changing. You try an AI tool, it helps a little, you move on.
Errol didn't stop. He did something different. He audited his entire week, found eight repeating tasks, and built templates for five of them.
His time savings went from 20 minutes per week to 10 hours per week. Same job title, same responsibilities, completely different experience of work.
I feel like I have a very new job. The same job, but I'm doing a different piece of work. Less boring stuff, more thinking.
This is the difference between people who try AI workflows and people who build systems. Here's exactly what Errol did and how you can replicate it.
Starting Point: One Workflow
Like most people who start using AI for work, Errol began with the obvious task. Stakeholder updates are weekly, they follow a predictable format, and nobody likes writing them. Perfect candidate.
He built a simple prompt template. Paste in the week's raw updates. Get a structured stakeholder email with milestones, blockers, and next week's plan. Twenty minutes saved per week.
This is where most people stop. The workflow works, they use it a couple times, and the novelty wears off. They go back to manual work because "AI isn't that useful anyway."
The problem isn't AI. The problem is stopping at one.
The Audit That Changed Everything
Errol went back to his calendar and did a full audit. He listed every task he does more than once per week. Not the big strategic stuff. The repetitive operational work that eats up his days.
He found eight tasks:
- Weekly status reports
- Meeting notes
- PRD reviews
- Sprint planning preparation
- Stakeholder emails
- And three more he identified as partially automatable
For each task, he asked the same question: "Does this follow a predictable pattern even when the content changes?" All five of those top tasks did.
Building Five Templates
He built a prompt template for each of the five tasks. Not complicated templates. Role-based prompts with clear inputs and expected outputs.
For meeting notes: paste in raw notes, get structured summary with action items and decisions.
For PRD reviews: paste in a PRD, get a structured review with gaps, inconsistencies, and suggested improvements.
For sprint planning: paste in the team's completed work and upcoming backlog, get a prep document with capacity analysis and recommendations.
For stakeholder emails: paste in project updates, get a professional email formatted for the recipient's role.
Each template took maybe 10 minutes to write and refine. With AI helping write the initial versions, it was even faster.
The Numbers
Here's what changed:
- Before: 1 workflow, 20 minutes saved per week
- After: 5 workflows, 10 hours saved per week
- Time to build: About 2 hours total (audit + templates + testing)
- Ongoing maintenance: Near zero. Templates either work or get tweaked.
That's a 30x return on his initial 20-minute weekly savings. And the 2-hour investment pays dividends every single week.
But the real payoff isn't the time savings. It's what he did with the recovered time. Strategic thinking. Cross-functional collaboration. Work that actually requires a human brain instead of a template.
The Difference: System vs. Shortcut
What separated Errol from people who try AI and quit? He built a system instead of finding a shortcut.
A shortcut is: "I found a way to do this one thing faster." You use it, it helps, you don't think about it again.
A system is: "I have a process for finding and automating repeating tasks. I do it monthly. My library grows. My free time grows with it."
Errol's system has three parts:
- Monthly audit: 15 minutes. Open calendar, list repeating tasks, score for automation potential.
- Template building: 30 minutes for top 3 tasks (less with AI help). Write role-based prompt templates.
- Weekly use: Use the templates. Save what works. Refine what doesn't.
Total monthly investment: about an hour. Ongoing return: 10 hours per week.
Why This Works for Any Role
Errol is a PM, but this pattern works for any knowledge worker. Developers can automate code reviews, PR descriptions, and technical documentation. Designers can automate design briefs, component specs, and style guide updates. Lecturers can automate lesson plans, flashcards, and grading rubrics. Students can automate study guides, reading summaries, and interview prep.
The roles change. The tasks change. The system doesn't. Audit your week. Pick the top tasks. Build templates. Use them. Repeat monthly.
The Lesson
One workflow saving 20 minutes per week doesn't feel transformative. Five workflows saving 10 hours per week changes how you experience your job.
The gap between those two outcomes isn't AI capability. It's willingness to look beyond the obvious task and audit your entire work pattern. Most repeating work is invisible because you've always done it that way.
Make it visible. Write it down. Build templates. And don't stop at one.
If you want to try this yourself, we run free weekly workshops where we build workflows together, live. Join the next one.