Behind the Scenes
264,486 Emails.
One Claude Code Session.
On March 22, 2026, Glen downloaded his entire email history from three Google accounts and fed it to an AI coding agent. What came out the other side was the /story section of this website. This is how that happened.
264,486
Total Emails
23
Years
3
Accounts
1
Session
Step 1: Download Everything
Google Takeout. Three accounts. Export format: MBOX. Glen hit “Download” on each account and waited. The files landed at different sizes because each account lived a different life.
The idea was simple and slightly unhinged: what if you gave an AI your entire digital paper trail and asked it to tell your story? Not a curated highlights reel. Not a LinkedIn summary. The whole thing — every receipt, every argument, every 2 AM email to your lawyer.
The Personal Account
globalspeculation@gmail.com
144,750
2003 – 2026
Everything. The hedge fund. The GSE fight. The dating apps. The text messages forwarded to email. 23 years of Glen's digital life in one inbox.
30,574 emails from info@salesforce.com alone
The Corporate Account
Innovate Work Account
96,791
2014 – 2026
A decade of Salesforce development at Innovate Inc. Sprint planning, code reviews, deployment notifications, and 96,791 reasons Glen can navigate any org chart.
Peak year: 2024 with 18,203 emails
The Startup Account
Nimba Solutions Account
22,945
2021 – 2026
The consulting business. Watch a company explode in real-time: from 685 emails in Year 1 to 9,734 emails in Year 5. That growth curve tells the whole story.
14x email volume growth in 4 years
Combined total: 264,486 emails. If you printed them all single-sided, the stack would be roughly 88 feet tall. Nobody is going to read that. But an AI will.
Step 2: Feed It to Claude Code
Claude Code is a terminal-based AI agent. It reads files, writes code, runs commands, and commits to git. Glen pointed it at the MBOX exports and said something like “tell me my story.”
What happened next was genuinely interesting. The AI didn’t just count emails. It found patterns. It identified the people who showed up across all three accounts. It noticed that “Rop v. FHFA” appeared every single year from 2017 to 2023. It mapped the trajectory of Cloud Nimbus from a side project to a real business by watching the email volume hockey stick.
It found the chapters of a life that Glen hadn’t organized himself. The data did the organizing.
Email Volume by Year
All three accounts combined. 2026 data is Q1 only.
Step 3: The Patterns Emerge
When you analyze 264,486 emails, you stop seeing messages and start seeing a life. Here’s what stood out.
20+
People with 1,000+ emails each
The GSE investor network. Names that appear in every chapter of the Fannie Mae fight. These aren't contacts — they're co-combatants.
3–5
Parallel projects at all times
The data confirmed what everyone already suspected: Glen has never done just one thing. At any given point, there are 3 to 5 active threads running simultaneously.
7
Years of Rop v. FHFA emails
The case appears in the data from 2017 through 2023. Seven years of litigation, strategy, and hope. Every year, without fail.
30,574
Emails from Salesforce
info@salesforce.com wins the all-time sender championship by a landslide. Deployment notifications, Trailhead badges, org alerts, release notes. Not even close.
14x
Cloud Nimbus email growth
685 emails in 2021. 9,734 in 2025. That's what a consulting business scaling looks like in raw inbox data.
1
Session to build everything
Claude Code read the data, identified the patterns, found the people, mapped the timeline, and built the entire /story section. One session. March 22, 2026.
Step 4: Build the Story
Claude Code didn’t just analyze the data. It built the website. The entire /story section — the main biographical page, the chapter subpages, the people directory — was constructed from the patterns in the email data.
The process looked like this:
- 1
Ingest
Parse MBOX files. Extract sender, recipient, date, subject, and body text from every email.
- 2
Identify
Find recurring names, organizations, and topics. Map relationships by co-occurrence across threads.
- 3
Timeline
Plot major life events by detecting spikes in email volume and shifts in sender composition.
- 4
Verify
Cross-reference email patterns with public records — Seeking Alpha articles, court filings, LinkedIn history.
- 5
Write
Generate the narrative. Not a summary — a story told through the data, with real stats woven in.
- 6
Build
Write the Next.js pages, TypeScript components, JSON-LD schemas, and SEO metadata. Deploy via git push.
The meta part: This page you’re reading right now? Also built by Claude Code, in the same session, as the documentation for its own process. It’s turtles all the way down.
“Most people curate their story. I just downloaded mine. 264,486 emails, unfiltered, handed to an AI that doesn’t know how to be polite about patterns. Turns out the data tells a better story than I could.”
— Glen Bradford, March 22, 2026
Keep Exploring
The Full Story
The biography that came out of those 264,486 emails. The complete Glen Bradford story.
Read moreDirectoryThe People
The 20+ people who appeared across all three accounts. The co-characters in this story.
Read moreCase StudyBuilt with Claude Code
How this entire 790+ page website was built by an AI coding agent for $200/month.
Read more