Skip to content

Inkwell Tutorial: Processing Your First Podcast Episode

Time Required: 10 minutes Difficulty: Beginner Prerequisites: Python 3.10+, ffmpeg, API keys

What We'll Build

By the end of this tutorial, you'll have: - ✅ Inkwell installed and configured - ✅ A podcast feed added - ✅ Your first episode processed - ✅ Structured markdown notes in Obsidian - ✅ Understanding of costs and optimization

Step 1: Installation (3 minutes)

1.1 Install uv

curl -LsSf https://astral.sh/uv/install.sh | sh

1.2 Install Inkwell

uv tool install inkwell-cli

1.3 Install ffmpeg

macOS: 

brew install ffmpeg

Ubuntu/Debian: 

sudo apt-get install ffmpeg

1.4 Verify Installation

inkwell version
# Output: Inkwell CLI v0.20.0

Step 2: API Keys (2 minutes)

2.1 Get Google AI API Key

  1. Go to https://aistudio.google.com/api-keys
  2. Click "Create API Key"
  3. Select a project (or create new)
  4. Copy your API key

Use the Inkwell CLI to store your API key in the config file:

inkwell config set transcription.api_key "your-google-ai-api-key-here"

This stores the key in Inkwell's config file, scoped to this tool only. Gemini extraction uses this key too unless you set extraction.gemini_api_key separately.

Alternative: Environment Variables

If you prefer environment variables (useful if you use the same key across multiple tools):

# Temporary (current session)
export GOOGLE_API_KEY="your-google-ai-api-key-here"

# Permanent (add to ~/.zshrc or ~/.bashrc)
echo 'export GOOGLE_API_KEY="your-google-ai-api-key-here"' >> ~/.zshrc
source ~/.zshrc

2.3 Set Output Directory (Optional)

By default, Inkwell saves processed episodes to ~/podcasts. To change this:

inkwell config set default_output_dir ~/my-notes/podcasts

You can also specify a different directory per-command using the --output flag.

2.4 Verify Configuration

inkwell config show

Step 3: Add Your First Podcast (1 minute)

Let's add Lenny's Podcast (a popular product and growth podcast):

inkwell add \
  "https://api.substack.com/feed/podcast/10845.rss" \
  --feed-name lennys \
  --category tech

Output: 

✓ Feed added: lennys
  URL: https://api.substack.com/feed/podcast/10845.rss
  Category: tech

3.1 List Available Episodes

inkwell list

Output:

inkwell list output

Step 4: Process Your First Episode (3 minutes)

4.1 Fetch Latest Episode

inkwell fetch lennys --latest

What Happens

  1. Downloads episode metadata
  2. Checks for YouTube transcript (free!)
  3. Extracts content using Gemini
  4. Generates markdown with wikilinks and tags
  5. Tracks costs

Output: 

Processing: Building a Growth Team

Transcription: YouTube API (free) ✓
Extraction:    Gemini Flash      ✓

Templates:     4 (summary, quotes, key-concepts, tools-mentioned)
Cost:          $0.0055
Output:        ./output/lennys-2025-11-10-building-a-growth-team/

✓ Complete!

4.2 Explore the Output

cd output/lennys-2025-11-10-building-a-growth-team
ls -lh

Files Created: 

.metadata.yaml          # Episode metadata and costs
summary.md              # AI-generated summary
quotes.md               # Notable quotes
key-concepts.md         # Key concepts explained
tools-mentioned.md      # Tools and resources

4.3 View a File

cat summary.md

Example Output: 

---
title: Summary
podcast: Lenny's Podcast
episode: Building a Growth Team
episode_date: 2025-11-10
duration_minutes: 45
rating: 4
status: inbox
tags: [podcast, growth, product-management, startups]
has_wikilinks: true
cost_usd: 0.0055
---

# Building a Growth Team

## Overview

In this episode, Lenny discusses how to build and structure a growth team with a seasoned growth leader. They cover [[Growth Loops]], [[A/B Testing]], [[Product-Led Growth]], and team composition strategies.

## Key Takeaways

- **[[Growth Teams]]** should be cross-functional with engineering, design, and data
- **[[Experimentation Velocity]]** is more important than individual experiment wins
- **[[North Star Metrics]]** help align the team on what matters
- Start with [[Activation]] before focusing on [[Retention]]

## Practical Applications

The guest shares how they structured growth at their company and common pitfalls to avoid when scaling.

## Related Topics

- #growth #product-management #startups
- [[Product-Led Growth]] [[Experimentation]] [[User Activation]]

Step 5: Open in Obsidian (1 minute)

5.1 Configure Obsidian Vault

  1. Open Obsidian
  2. Open your vault (or create new one)
  3. Copy episode folder to vault:
cp -r output/lennys-2025-11-10-building-a-growth-team ~/ObsidianVault/podcasts/

5.2 Navigate to Notes

In Obsidian: 1. Open Files pane (Ctrl/Cmd + O) 2. Navigate to podcasts/lennys-2025-11-10-building-a-growth-team/ 3. Open summary.md

Click on any [[Growth Teams]] link - Obsidian will: - Create a backlink - Allow you to create a new note for that topic - Show related notes

5.4 Use Dataview

Create a new note called "Podcast Dashboard":

# Podcast Dashboard

## Recent Episodes
```dataview
TABLE episode, rating, duration_minutes, cost_usd
FROM "podcasts"
WHERE template = "obsidian-note"
SORT episode_date DESC
LIMIT 10

By Topic

TABLE episode, podcast
FROM "podcasts"
WHERE contains(topics, "growth")
SORT episode_date DESC

## Step 6: Check Costs (30 seconds)

```bash
inkwell costs

Output: 

┌ Overall ─────────────────────┐
│ Total Operations:  1          │
│ Total Cost:        $0.0055    │
│                               │
│ By Provider:                  │
│   gemini    $0.0055           │
│                               │
│ By Operation:                 │
│   extraction    $0.0055       │
└───────────────────────────────┘

Cost Breakdown: - Transcription: $0.00 (YouTube was free!) - Extraction: $0.0055 (Gemini Flash) - Total: $0.0055 ✨

Step 7: Process with Interview (Optional, +5 minutes)

Try interactive interview mode to capture your insights:

inkwell fetch lennys --latest --interview

Interactive Session: 

Episode processed! Starting interview...

Q: What was your main takeaway from this episode?
A: Growth teams need to be cross-functional from day one.

Q: How does this relate to your current work?
A: We're building our first growth team. Need to involve engineering earlier.

Q: What will you do differently based on this?
A: Set up a weekly experimentation review with the whole team.

Q: Any resources to follow up on?
A: Read Lenny's newsletter on growth loops, check out Reforge courses.

✓ Interview complete! Saved to my-notes.md

What You Learned

✅ Installation: Set up Inkwell and dependencies ✅ Configuration: API keys and basic settings ✅ Feed Management: Add and list podcast feeds ✅ Episode Processing: Fetch and extract content ✅ Obsidian Integration: Wikilinks, tags, Dataview ✅ Cost Tracking: Monitor and optimize spending ✅ Interview Mode: Capture personal insights

Next Steps

Process More Episodes

# Process latest 5 episodes
inkwell fetch lennys --count 5

# Process specific episode
inkwell fetch lennys --episode 789

Try Different Podcasts

# Add more podcasts
inkwell add "https://feeds.simplecast.com/54nAGcIl" --feed-name syntax

# Process
inkwell fetch syntax --latest

Customize Configuration

# Edit config
inkwell config edit

# Try different settings:
# - Change output directory
# - Enable/disable specific templates
# - Adjust tag limits
# - Configure Dataview fields

Explore Advanced Features

  • Custom Templates: Create your own extraction templates
  • Batch Processing: Process multiple episodes automatically
  • Cost Optimization: Strategies to minimize API costs
  • Dataview Queries: Advanced Obsidian queries

Troubleshooting

"No API key found"

Make sure you exported the API key: 

export GOOGLE_API_KEY="your-key"
echo $GOOGLE_API_KEY  # Should print your key

"ffmpeg not found"

Install ffmpeg: 

# macOS
brew install ffmpeg

# Ubuntu
sudo apt-get install ffmpeg

"Episode already exists"

Use --overwrite to replace: 

inkwell fetch syntax --latest --overwrite

High Costs

Check if YouTube transcript is available (free): 

inkwell costs --recent 5

If using Gemini transcription frequently, consider: - Caching transcripts - Batch processing - Setting cost limits

Summary

You've learned how to: 1. Install and configure Inkwell 2. Add podcast feeds 3. Process episodes automatically 4. Generate Obsidian-ready notes 5. Track costs 6. Use interview mode

Typical Workflow: 

# Daily routine (2 minutes)
inkwell fetch lennys --latest
inkwell fetch syntax --latest
inkwell costs

# Weekly review (5 minutes)
# Open Obsidian, review notes, update ratings

Next Steps: - Processing Guide - Templates Reference

Feedback

Questions or issues? - GitHub Issues: https://github.com/chekos/inkwell-cli/issues - Documentation: https://chekos.github.io/inkwell-cli/

Happy note-taking! 🎧📝