Claude Usage — Token Analytics

1n2.org × Claude · Feb 7 – Apr 7, 2026 · 60 days of collaboration

💬 Watch Our Conversation

An iMessage-style animated replay of every prompt and response from our 28-day collaboration — from setting up a Mac Mini to building 40 projects, a data warehouse, an animated Bitcoin train with Mad Bitcoins conductor, and this very page. Messages appear with typing indicators and result cards showing what we built together.

▶ Play Conversation Animation

429+

Chat Sessions

~36M+

Est. Tokens

148

Projects Deployed

Days Active

Scheduled Tasks

304

Sessions (Last 7d)

~$2/mo

API Cost

95%

Sub / Local Runs

Token estimate based on subscription + local inference. 95% of sessions run on Claude Pro subscription or local Gemma 4 — only ~$2/mo API spend.

Analysis & Insights

Productivity Pattern: Phases 1–4 (Feb 7–Mar 8) built 40 projects in 30 days. Phase 5 (Mar 8–Apr 7) exploded to 108 new deployed sites in 30 days — 393 Dispatch/Cowork sessions with 304 in the last 7 days alone.

Token Distribution: Phases 1–4 used ~8.5M tokens across 36 sessions. Phase 5 added ~27.4M across 393 sessions — encyclopedias alone burned ~10M. The shift to subscription and local inference means 95% of runs cost nothing beyond the monthly Pro fee.

Encyclopedia Era: Phase 5 pivoted to mass-producing topic encyclopedias — Oakland A's, Beatles, Pete Rose, Jeopardy, Mets, shoes, malls, video games, fountain soda — each a standalone deployed site on 1n2.org.

Bot Evolution: "Number One" AI bot migrated from Anthropic Claude API → Ollama qwen → Ollama gemma → llama.cpp Gemma 4, achieving near-zero marginal cost for local inference on Mac Mini M4.

Complete Chronological Timeline

Every chat session and project in order — sessions sorted by date, projects grouped with their session

Date	Session / Project	Category	Est. Tokens	What Was Built
📅 PHASE 1 — Foundation (Feb 7–11)
Feb 7	GLM-4.flash model loading error in LMStudio	AI	~50K	Debugging local AI model loading on Mac Mini M4
Feb 7	Mac mini M4 GPU compatibility with LMStudio	AI	~50K	GPU setup for local inference, model selection
Feb 9	MediaLog project ⭐	App	~500K	Full PHP/MySQL app: 10 pages, visualizations, lists, export, 782 books tracked
Feb 9	→ Case Studies	App	—	Portfolio case studies site documenting MediaLog build
Feb 10	Building a topic-based Twitter reader	App	~300K	Tweetster: Nitter backend, topic tracking, harvest system
Feb 10	Creating a new tweetster project directory	Infra	~200K	Tweetster dark mode, read tracking, fetch button patches
Feb 10	→ MadPatrol	Game	—	Moon Patrol: Bitcoin Battle retro game
Feb 10	Configuring OpenClaw with Sonnet	AI	~50K	OpenClaw gateway config, model switching
Feb 10	Openclaw command for logs and updates	AI	~50K	Log viewing, update commands
Feb 11	Deploying local site to droplet	Infra	~200K	DigitalOcean droplet, Nginx, SSL, SSH deployment
Feb 11	Finding short domain names	Infra	~50K	Domain research → chose 1n2.org
Feb 11	Local app access and file editing	Infra	~50K	Local dev environment, PHP server, file access
Feb 11	→ Checklister	App	—	Task management app
Feb 11	→ AI Workflows	App	—	Workflow documentation site
🃏 PHASE 2 — Curio Cards Ecosystem (Feb 13–19)
Feb 13	Curio Cards NFT tracking and charting website	NFT	~400K	CurioCharts: React dashboard, OpenSea data, 31 cards tracked
Feb 14	CurioCharts program status check	NFT	~150K	Debugging and completing CurioCharts features
Feb 14	Building Curio Archive site ⭐	NFT	~400K	Encyclopedic history of Curio Cards with full card database
Feb 15	Connecting a droplet to a domain name	Infra	~200K	DNS config, Nginx vhosts, SSL certs for 1n2.org
Feb 15	→ Homepage v2	App	—	1n2.org redesign: product showcase, dev timeline, icons
Feb 17	→ Thomas Hunt Films	App	—	Film portfolio: 38 videos, 5.5M views, 7 categories
Feb 17	Bulk follow accounts on X	Infra	~80K	Twitter automation for following crypto accounts
Feb 18	Finishing CurioCharts project	NFT	~300K	Polish, deploy, final features for CurioCharts
Feb 18	Update 1n2.org droplet and mirror host files	Infra	~200K	Server mirroring, file sync, deployment scripts
Feb 18	→ Curio Terminal	NFT	—	Professional analytics dashboard for Curio Cards
Feb 18	→ Curio Atlas	NFT	—	Network & ecosystem mapping tool
Feb 18	→ Curio Data API	NFT	—	Live market data service
Feb 18	→ Curio Review	NFT	—	Weekly AI-generated review essays
Feb 19	→ CurioQuant	NFT	—	AI trading signals platform
Feb 19	→ CurioHub	NFT	—	Central Curio Cards ecosystem hub
Feb 19	→ Daily Logs	Data	—	AI + Human collaboration journal
⚙️ PHASE 3 — AI Pipeline & Config (Mar 4–5)
Mar 4	Switching AI models with limited credits	AI	~80K	Model fallback chains, cost optimization
Mar 4	PHP server startup configuration	Infra	~80K	@reboot cron, localhost:8000 PHP server
Mar 4	Switching Openclaw to Kimi 2.5	AI	~80K	Testing Kimi 2.5 as alternative to Claude
Mar 5	Configuring OPENCLAW after returning home	AI	~150K	Full OpenClaw reconfiguration: Ollama, Telegram, models
Mar 5	Local PHP server startup configuration	Infra	~100K	Persistent PHP server via LaunchAgent
Mar 5	Openclaw command for live log viewing	AI	~100K	Real-time log tailing, debugging tools
Mar 5	Telegram connectivity	Infra	~100K	Telegram bot integration with OpenClaw
🚀 PHASE 4 — Full-Stack Sprint (Mar 5–8)
Mar 5	→ MB Games	Game	—	Mad Bitcoins Retro Game Lab (collection of games)
Mar 5	→ Newszor	Data	—	Google News aggregator, 32 articles/day, topic scraping
Mar 5	→ MadWikipedia	App	—	Encyclopedia of Mad Bitcoins / WCN / Curio Cards
Mar 5	→ Curio-DID	Data	—	Causal analysis tool with academic research paper
Mar 6	Daily THunt video pipeline automation ⭐	AI	~800K	Text extraction → TTS (Voicebox) → Hedra video → Telegram publish
Mar 6	Building Curio Terminal backend ⭐	NFT	~1.5M	Full backend scripts, MVP frontend, analytics dashboard
Mar 6	→ Data Labs	Data	—	Central SQLite warehouse: 16 tables, 176K records, 33MB
Mar 6	→ Commentzor	Data	—	YouTube comments aggregator (38K+ comments)
Mar 6	→ Bitcoin Trains	Game	—	Full BTC history replay: animated train, Mad Bitcoins conductor
Mar 6	→ Facebooker	Data	—	Facebook page digest tool
Mar 6	→ Health Check	Infra	—	System health monitoring: DB tables, site checks, cron status
Mar 6	→ Mad Ninja	Game	—	Shadow of the Blockchain (ninja platformer game)
Mar 7	YouTube comment database aggregator ⭐	Data	~1.2M	YouTube API: 4 channels, 5,439 videos, 38,597 comments, Alchemy NFT data
Mar 7	Setting up cron jobs ⭐	Infra	~700K	14-job cron pipeline (master-run.sh), Letterboxd/Goodreads import
Mar 7	→ Dashboarder	App	—	6-tab dark dashboard: Briefing, Personal, Media, Comments, Theater, Museums
Mar 7	→ Signal Studio	NFT	—	Curio trading signal generator
Mar 7	→ Curio Oracle	NFT	—	AI-powered Curio Cards price oracle
Mar 8	→ Agent Hangar	AI	—	AI agent management hub
Mar 8	→ Briefsmith	App	—	Content briefing tool
Mar 8	→ Claude Usage	Data	—	This page — token analytics & collaboration history
Mar 8	→ Clip Arcade	Game	—	Video clip collection
Mar 8	→ Codex Night Shift	AI	—	Overnight automation hub
Mar 8	→ Trend Weaver	Data	—	Trend analysis tool
🔥 PHASE 5 — Dispatch Era (Mar 8 – Apr 7) · 393 sessions · 108 new sites
Mar 8–15	Oakland A's Encyclopedia ⭐	App	~1.0M	Comprehensive Oakland A's history site — rosters, seasons, stadiums
Mar 8–15	→ A's Hats	App	~250K	Visual encyclopedia of Oakland A's hat designs through the years
Mar 8–15	→ Beatles Encyclopedia	App	~900K	Comprehensive Beatles discography, history, and cultural impact site
Mar 8–15	→ 1986 Mets	App	~800K	Deep dive into the 1986 Mets championship season
Mar 8–15	→ Pete Rose	App	~800K	Pete Rose career encyclopedia — stats, timeline, controversies
Mar 8–15	→ Jeopardy Encyclopedia	App	~700K	Jeopardy! show history, champions, categories, and records
Mar 15–22	Best Picture Tracker ⭐	App	~500K	Complete Best Picture history with TMDB integration
Mar 15–22	→ Oscar Quiz	Game	~1.2M	Interactive Oscar quiz — IndexedDB storage, 351 TMDB movie posters, batch API fetches
Mar 15–22	→ Flavors Encyclopedia	App	~500K	Ice cream and flavor history encyclopedia
Mar 15–22	→ Fountain Soda	App	~400K	History and culture of fountain sodas and soda fountains
Mar 15–22	→ Borg Drinks	App	~400K	Encyclopedia of Borg drink recipes and culture
Mar 15–22	→ Shoes Encyclopedia	App	~500K	Sneaker and footwear history encyclopedia
Mar 15–22	→ Malls Encyclopedia	App	~600K	American mall culture, history, and architecture
Mar 15–22	→ Video Game History	App	~700K	Comprehensive video game history from Pong to present
Mar 22–29	Palm Connection	App	~400K	Palm Computing nostalgia and history site
Mar 22–29	→ Girlfriend Vision	App	~350K	Entertainment/dating culture analysis tool
Mar 22–29	→ My Melody	App	~300K	Sanrio My Melody fan encyclopedia
Mar 22–29	→ Smooth Criminal Soundboard	Game	~250K	Interactive Michael Jackson soundboard with audio clips
Mar 22–29	→ LCARS Screensaver	App	~200K	Star Trek LCARS-style animated screensaver
Mar 22–29	→ Umbrella Revolution	App	~600K	WCN archive site documenting the Hong Kong Umbrella Movement
Mar 22–29	→ Deals (CheapCharts variant)	Data	~350K	Digital media deal tracker, CheapCharts-inspired
Mar 29–Apr 4	Media Manager ⭐	App	~1.0M	5 Python scripts + 8-tab web UI for Plex media management
Mar 29–Apr 4	→ Tom & Jerry Analysis	Data	~400K	Tom & Jerry cartoon violence analysis — episode-by-episode data
Mar 29–Apr 4	→ Curio Cards Market Updates	NFT	~300K	Updated market analysis and pricing data for Curio Cards
Mar 29–Apr 4	→ WCN Podcast Hub	Data	~400K	World Crypto Network podcast aggregation hub
Mar 29–Apr 4	→ WCN Tracker	Data	~350K	World Crypto Network show and episode tracker
Mar 29–Apr 4	→ Thomas Hunt History	App	~500K	Thomas Hunt biographical and career history site
Mar 29–Apr 4	→ Thomas Hunt Films (v2)	App	~200K	Updated film portfolio and showcase
Apr 1–7	Podcast Ops Pipeline ⭐	AI	~400K	ai-marketing-skills podcast operations and automation pipeline
Apr 1–7	→ YouTube Shorts Pipeline	AI	~100K	YouTube Shorts generation pipeline (cloned, not yet configured)
Apr 1–7	→ Godin Fame System	App	~300K	Seth Godin-inspired fame/branding system site
Apr 1–7	→ Number One Tools Suite	AI	~200K	bin/ scripts for Number One bot operations and management
Apr 1–7	→ AI Usage Dashboard (v2)	Data	~300K	This page — updated with Phase 5 data through Apr 7
Mar 8–Apr 7	36 Scheduled Tasks configured ⭐	AI	~7.5M	20+ daily tasks × 30 days × ~50K/session — automated site builds, data pipelines, monitoring
Mar 8–Apr 7	Number One bot migration ⭐	AI	~500K	Claude API → Gemini → Ollama qwen → Ollama gemma → llama.cpp Gemma 4
Mar 8–Apr 7	~75 additional deployed sites	App	~4.0M	Numerous smaller encyclopedia, tool, and utility sites (~50K each avg)
Phase 5 Subtotal			~27.4M	Encyclopedias ~10M · Scheduled tasks ~7.5M · Apps ~5M · AI/Infra ~2.5M · Data ~2.4M

Token Usage by Category

Category	Projects	Est. Tokens
Apps / Encyclopedias	95+	~16.5M
AI / Automation	20+	~9.5M
Data / Pipeline	18	~4.5M
NFT / Curio	14	~3.0M
Infrastructure	12	~2.5M
Games / Interactive	8	~2.0M
Total	148+	~35.9M

Phase	Duration	Sessions	Tokens
Phase 1 — Foundation	Feb 7–11 (5d)	14	~1.5M
Phase 2 — Curio Ecosystem	Feb 13–19 (7d)	12	~1.7M
Gap — Feb 20–Mar 3	12 days	—	—
Phase 3 — AI Pipeline	Mar 4–5 (2d)	7	~0.7M
Phase 4 — Full Sprint	Mar 5–8 (4d)	3	~4.6M
Phase 5 — Dispatch Era	Mar 8–Apr 7 (30d)	393	~27.4M
Grand Total	Feb 7–Apr 7 (60d)	429+	~35.9M

Infrastructure Stack

Component	Details
Server	DigitalOcean Droplet · 157.245.186.58 · Ubuntu · Nginx + SSL
Domain	1n2.org — 148 deployed projects (subdirectories)
Databases	SQLite (Data Labs: 16 tables, 176K records, 33MB) + MySQL (MediaLog) + IndexedDB (Oscar Quiz: 351 posters)
Cron Pipeline	14 daily jobs via master-run.sh + 36 Dispatch/Cowork scheduled tasks (20+ daily)
APIs	YouTube Data v3, Alchemy (Ethereum), CoinGecko, TMDB (351 movie posters), Google News RSS, Nitter
AI Stack (Cloud)	Claude Pro subscription (Dispatch/Cowork, primary builder) · Gemini API (brief stint)
AI Stack (Local)	llama.cpp + Gemma 4 on Mac Mini M4 (Number One bot) · Ollama (qwen, gemma — deprecated)
Orchestration	Claude Dispatch + Cowork (393 sessions in Phase 5) · 36 scheduled tasks · rsync deploys
Local Dev	Mac Mini M4 · PHP 8.3 · Python 3.12 · Node.js · 5 Plex Python scripts · localhost:8000
Deployment	rsync to droplet via SSH (ed25519) — automated in cron + Dispatch scheduled tasks

Key Metrics

Phase 5 velocity: 393 sessions in 30 days (13.1/day avg), peaking at 304 sessions in the last 7 days alone (43.4/day). That's a 12x increase over Phase 1–4's average of 1.2 sessions/day.

Deployment explosion: Went from 40 deployed sites to 148 — a 3.7x increase. New sites span sports encyclopedias, food/drink culture, entertainment, media tools, and automation pipelines.

Cost optimization: Shifted from Anthropic API (~$0.15/session) to Claude Pro subscription + local Gemma 4 inference. Result: ~$2/mo API cost, 95% of runs on subscription or local hardware.

Automation at scale: 36 configured scheduled tasks (20+ running daily) via Dispatch/Cowork. The system now builds, deploys, and maintains sites with minimal human intervention.

Phase 5 Token Breakdown

Where the ~27.4M tokens went — encyclopedias dominate, scheduled tasks are the hidden cost

Phase 5 Highlights

Media Manager: Full Plex management suite — 5 Python scripts + 8-tab web UI for library management, metadata editing, and media organization.

Oscar Quiz: Interactive Academy Awards quiz with IndexedDB persistence and 351 movie posters fetched from TMDB API. Largest client-side data store in the ecosystem.

Number One bot: Personal AI assistant migrated through 4 backends in 30 days — Anthropic Claude → Gemini → Ollama (qwen/gemma) → llama.cpp Gemma 4, now running fully local.

Encyclopedia factory: Developed a repeatable pattern for deploying topic-specific encyclopedias — Oakland A's, Beatles, Mets, Pete Rose, Jeopardy, shoes, malls, video games, fountain soda, and more.

Where Phase 5 Tokens Went

Biggest burner — Encyclopedias (~10M): 30+ topic encyclopedias at ~500K–1M each dominated Phase 5 spending. Oakland A's (1.0M), Beatles (900K), Mets (800K), Pete Rose (800K), Video Game History (700K), and Jeopardy (700K) were the heaviest. Even "smaller" encyclopedias like Shoes, Malls, and Flavors each consumed ~400–600K tokens due to multi-page content, data formatting, and iterative refinement.

Hidden cost — Scheduled Tasks (~7.5M): 20+ daily automated tasks running ~150 times across 30 days. Each session burns ~50K tokens for site rebuilds, data pipeline runs, Newszor updates, market reports, and health checks. Low per-session cost, massive aggregate — 27% of Phase 5 total.

Interactive apps (~2.2M): Oscar Quiz (1.2M) and Media Manager (1.0M) were the most token-intensive single projects — both required complex state management, API integrations, and iterative UI work across multiple sessions.

Long tail (~4M): ~75 additional smaller deployed sites averaged ~50K each. These include utility pages, micro-tools, and content sites that individually cost little but add up in aggregate to nearly 15% of Phase 5 spending.

1n2.org · Dashboarder · Bitcoin Trains · Data Labs · Health · Built with Claude · Updated April 7, 2026