Understand the context
Explore repositories, develop test ideas, and investigate failures with AI assistance. Product risks and existing behavior guide the questions.
Back to portfolio
Practical AI.
AI-assisted QA
Practical AI.
Verified results.
Daily use of Claude Code and Codex for repository analysis, test design, automation code, debugging, and documentation.
Build and improve
Draft and maintain test code, refactor shared utilities, and prepare documentation. Use agents to support repeatable QA work and problem analysis.
Check the result
Review proposed changes, run relevant checks, and verify behavior manually where needed. Technical judgment stays with the engineer.
ChatGPT / Codex activity
AI, in numbers.
Token usage and activity from my ChatGPT / Codex profile.
- Lifetime tokens
- 2,057,022,996
- Peak tokens in a day
- 264,324,130
- Longest task
- 17 hr 40 min
- Longest streak
- 31 days
- Current streak
- 8 days
Token activity by month
The 12 most recent calendar months. Bars compare reported token totals; the current month is incomplete.
- Current month · partial
- 71,011,810
- Days with reported activity: 2
- 500,484,534
- Days with reported activity: 25
- 1,295,095,542
- Days with reported activity: 30
- 73,536,737
- Days with reported activity: 10
- 76,084,272
- Days with reported activity: 9
- 18,999,442
- Days with reported activity: 8
- 21,810,659
- Days with reported activity: 9
- 0
- Days with reported activity: 0
- 0
- Days with reported activity: 0
- 0
- Days with reported activity: 0
- 0
- Days with reported activity: 0
- 0
- Days with reported activity: 0
A dated snapshot of account activity, not a measure of productivity or work quality.
Source: ChatGPT / Codex profile · Data updated:
About these numbers
Summary values come from the profile service. The longest task is elapsed task time, not hours worked. Monthly totals sum only the daily token values returned by that service; months without returned values show zero reported tokens, not proof of inactivity. Monthly totals and lifetime tokens cover different periods. Missing metrics remain unavailable. Conversations, task names, private projects, and credentials are not published. No statistics for Claude or other providers are included.