Teresa Torres uses it. Carl Vellotti's tutorial is great. Lenny's newsletter featured it.

But I looked at the examples and honestly? It felt a bit underwhelming.

Then I actually started using it. Two weeks. 59 sessions. 200+ hours saved.

I promise you: this is as big a productivity hack as voice-to-text (and you know I'm a voice-to-text maxi!). Maybe bigger.

Here's what actually happened.

Unless you've been living under a rock, you've seen Claude Code mentioned everywhere lately.

Teresa Torres uses it for competitive analysis. Product podcasts won't shut up about it. Your Slack communities are buzzing. I featured the Peter Yang / Carl Vellotti comprehensive tutorial. Lenny Rachitsky had over 50 featured PM use cases.

But here's what I missed: why it is actually better than the usual desktop Claude or CGPT, specifically for product managers?

Not developers writing code. Not engineers debugging applications. Product managers doing actual PM work.

But I decided to bite the bullet and try it. And haven't looked back. At the time of writing this, I'd spent 15 days—08 November through to 22 November—using Claude Code across 59 documented sessions. Four workflow categories. Product content automation. Career interview prep. Personal productivity systems. Business consulting research.

Every single session timestamped. Every output documented. Every ROI calculation tracked (OK with some healthy assumptions).

The outcome? 200+ hours saved. Real workflows automated. Reusable systems built.

This isn't theoretical. This isn't a weekend experiment where I automated one CSV file and called it a victory. This is how I work now. Every day. Multiple sessions. Compounding returns.

And if you're a PM drowning in data cleanup, document production, research synthesis, or any form of "glue work" - you need to understand why this actually matters.

Successful AI transformation starts with deeply understanding your organization’s most critical use cases. We recommend this practical guide from You.com that walks through a proven framework to identify, prioritize, and document high-value AI opportunities.

In this AI Use Case Discovery Guide, you’ll learn how to:

Get the Guide.

Let's demystify this immediately.

Claude Code is a text window for commands. That's it. You type instructions, Claude reads your files, makes changes, runs tasks, and gives you results.

It's not a mysterious developer IDE. It's not intimidating terminal wizardry. If you can organise folders on your computer, you can use Claude Code.

Here's what makes it different: Claude has direct access to your local files (and more). Not copying-and-pasting into a browser chat window. Not uploading files one by one to Claude Projects. Claude Code reads, writes, searches, and executes real things for you on your actual computer.

As you are clearly aware, I run Product Tapas - a weekly newsletter for product managers covering tools, insights, and bite-sized podcast summaries (Pod Shots). I run it side of the desk.

My reality before Claude Code:

Manual work estimate: on average it had crept up to 8+ hours per week. Not just the writing. Certainly not strategy. But data hygiene, file organisation, searching, updating, ordering and content transformation.

Claude Code changed this completely.

My tools database? Claude audited 581 entries, found 107 issues, tested whether the app was still live, cleaned duplicates, standardised formats, and created reusable scripts - all in one sitting. 20+ hours of manual work done in 2 hours.

My Pod Shots? Claude re-catalogued 110 episodes, enriched missing metadata, identified gaps, and created a searchable taxonomy. 160+ hours of podcast content → 8 hours of reading. That's a 95% time reduction. More usefully, I can now search across the entire resource that's wiki-linked for trends, conflicts and just interesting shit!

Newsletter automation? Claude Code helped built a pipeline (actioned in N8N): Gmail inbox → Claude analysis → Obsidian vault → Notion database → Beehiiv draft. 5-10 hours saved every single week.

This isn't coding. This is work automation for non-technical PMs. This is the workflow automation stuff you can do to save time right now.

You've used Claude in your browser. Maybe you've tried Claude Projects for team collaboration. So why does Claude Code actually matter?

Claude in Browser (claude.ai):

Claude Projects:

Claude Code (Terminal):

I really like Teresa Torres' simple explanation of Claude Code's advantage through a three-layer memory architecture:

Why this matters for PMs:

Browser Claude forgets everything when you start a new chat. Projects help, but require manual file uploads. Plus I find I run out of context so fast, so have to keep starting new chats.

Claude Code reads your entire project folder automatically. Newsletter archive? Pod Shots database? Tools CSV? Meeting notes? All available instantly. No uploads. No context re-explaining.

Your .claude/CLAUDE.md isn't just code preferences. It's your team's institutional knowledge:

Every session starts with this context. No re-explaining. No tribal knowledge loss when someone's on holiday.

Use case: Analyse 5 competitors simultaneously.

Browser Claude: You'd paste Company A info → get analysis → paste Company B info → get analysis → repeat 5 times. 30-45 minutes minimum.

Claude Code: You give Claude 5 company folders (or CSV files, or research docs). Claude analyses all 5 in parallel, creates comparison matrix, identifies gaps, and outputs structured report. 5-10 minutes total.

That's a 6X speed improvement. Not because Claude is faster. Because Claude Code eliminates the copy-paste friction entirely.

For PMs doing customer research, competitor analysis, feature comparisons, or data-heavy work? This is the difference between spending 3 hours or 30 minutes.

Here's something most people don't know exists: Planning Mode.

Claude Code has this brilliant hidden feature. Before executing any task, you can tell Claude to think through it first. Claude spends 5 minutes planning, shows you exactly what it's about to do, and waits for your approval.

Why this matters: Claude is helpful. Sometimes too helpful. It wants to solve your problem immediately and execute fast.

Real example from my vault: I asked Claude to add 3 new tags to my Notion database of 111 podcast summaries. "Be super careful," I said. "This is a live database."

Claude didn't read the documentation about Notion's MCP limits. It sent an update with just the 3 new tags—and deleted all 247 existing tags in the process. 111 entries, instantly untagged. 2 hours of careful recovery work.

Planning Mode would've caught this. Claude would've said: "I'm going to replace all existing tags with these 3 new ones. Proceed?" I would've stopped it immediately.

The lesson: Speed without strategy is expensive chaos. Let Claude think first. Approve second. Execute third.

Carl Vellotti covers this in his tutorial - Planning Mode prevents expensive mistakes. (Read the Podshot here) .

Let's be honest: There is a LOT of glue work in PM work.

Glue work is the unglamorous stuff. Data cleanup. Meeting notes synthesis. Research aggregation. Content transformation. Tool evaluation. File organisation. Helping bridge between teams.

It's necessary. It's time-consuming. And it's not what you want to do.

Here's where Claude Code gets powerful: compounding systems.

You spend 15 minutes setting up a workflow once. Then you execute it in 1 minute forever.

Example: Interview Prep Reusable Template

I had 8 company interviews in 2 weeks. Each required:

Manual effort per interview: 3-4 hours.

Claude Code workflow:

Result: 45 minutes per interview. 2-3 hours saved each time. 8 interviews = 16-24 hours returned to actual interview practice.

And the template? Reusable forever. Every future interview uses the same 45-minute workflow.

Product Tapas Content (51% of my Claude Code sessions):

Career & Interview Prep (22% of sessions):

Personal Productivity (14% of sessions):

Business Consulting (13% of sessions):

ROI Preview: 200+ hours saved in 15 days. That's 67-92% net positive ROI already—and I'm just getting started.

(Part 2 will break down every single workflow with receipts, dashboard visualisations, and reusable templates you can steal. For now, just know: this is real, this is replicable, and this is for PMs.)

Quick edit on the above — because I just couldn't resist. I got Claude Code to build me an interactive Python dashboard that analyses my plans, looks at time spent, breaks down the splits across different work I've done, and spits out a dashboard on my local machine.

The one above is the lovely Nano Banana image — much cleaner — but below is my dashboard as of today. I've gone from 59 plans to 87 in the last four days. I'm clearly spending too much time in Claude Code; I'm addicted.

Imagine reclaiming 10-20 hours per week. What would you do with that time?

Claude Code doesn't eliminate all glue work. But it automates the mechanical, repetitive, data-heavy bits.

The 15-minute daily note generation? Automated. The 3-hour data cleanup? Automated. The 2-hour research synthesis? Automated.

You're still the strategist. Claude Code handles the grunt work.

"But I'm not a developer. I don't know terminal commands. This sounds intimidating."

I get it. I felt the same way on 08 November.

Here's the reality: You need to master 3-4 commands. That's it.

Prerequisites:

Installation (10 minutes):

That's it. You're done. No coding. No configuration files. No developer wizardry.

Pick one messy data problem. A CSV file. A spreadsheet. A folder of documents. Something with duplicates, missing data, or inconsistent formatting.

Give Claude Code this prompt:

"Analyse this CSV file. Tell me:

Claude will:

You've just automated 2-4 hours of manual Excel work in 2 minutes.

Week 1: Basic file operations (read, search, analyse, create summaries)

Week 2: Transformations and automations (cleanup, format changes, batch operations)

Week 3: Multi-agent orchestration and complex systems (parallel research, multi-step workflows)

This progression is non-linear. Some will master Claude Code in 3 days. Some 3 weeks. It depends on:

By Week 2-3, you'll want to build custom slash commands. These turn 30-minute tasks into 2-minute shortcuts.

Mine:

And skills (reusable workflow templates):

Not coding. Just: "Hey Claude, when I type /daily-note, do these 8 steps automatically."

Permission-giving truth: You don't need to understand how Claude works under the hood. You don't need to learn Python or JavaScript. You don't need to become a developer.

You just need to describe what you want in plain English. Claude does the rest.

Part 1 explained why Claude Code matters for PMs. Now you understand:

Part 2 drops next week: "How I Saved 200+ Hours with Claude Code: Real PM Workflows"

The full breakdown:

Plus: The reusable templates and workflows you can steal immediately.

In the meantime: Try the CSV cleanup exercise above. Let me know your results. I'll answer every question.

✓ Claude Code = AI assistant with direct file access (game-changer for PMs drowning in data work)

✓ Non-technical PMs can use this (organise folders = sufficient skill, no coding required)

✓ Planning Mode prevents disasters (5 min thinking saves 2 hours fixing, use it)

✓ Compounding systems beat one-off tasks (invest 15 min once, benefit 1 min forever)

✓ Real PM workflows exist (data cleanup, content automation, research synthesis, interview prep)

✓ 200+ hours ROI possible in 15 days (proof coming in Part 2 with dashboard + receipts)

What PM workflow is your biggest time-sink right now?

Data cleanup? Research synthesis? Meeting notes? Content production?

Let me know. Part 2 might just have your answer.

And if you try the CSV cleanup exercise above—share your before/after. Let's learn together.

This Series:

Referenced sources:

That’s a wrap.

Hope you loved this. Can't wait to share the rest with you!

Please share and let me know what you would like to see more or less of so I can continue to improve Product Tapas. 🚀👋

Alastair 🍽️.