tokenpayback

Your AI activity ledger — every agent, every activity, 100% local.

loading…

Spent on AI

—

all sessions

Human-equivalent value

—

low — high

Time AI saved

—

vs human professionals

Overall ROI (human cost / AI cost)

—

What your AI did this period

Every session is categorized. Asking questions counts. Personal tasks count. Code shipped is just one of many ways your tokens turned into something.

Across your AI agents

tokenpayback auto-detects Claude Code, Codex CLI, Hermes, OpenClaw, OpenHuman, and Cursor on your machine.

Agent	Sessions	$ spent	est. value	ROI

Last 4 weeks (engineering output)

Week	Cost	Value	ROI	PRs	Commits	+Lines	Reverts

Recent sessions — with human-equivalent valuation

Each session lists the closest professional role that would normally do this work, the estimated human time, and the value at that role's market rate. Quality factor reflects how usable the AI output was. Nothing leaves your machine.

Date	Agent	Via	Category	Project	Equivalent role	Human time	Quality	$ AI cost	$ Human-equiv	ROI

How we estimate value

Per session, the LLM picks the closest professional role that would do this work and estimates hours + market rate (anchored to your benchmark: us-west / senior). Value = hours × rate × quality_factor. Quality factor: full-replacement 1.0 · with-edits 0.7 · draft-only 0.4 · failed 0.0.

Estimates are uncertain by design — that's why every number is a range, not a single value. Tune the benchmark in ~/.tokenpayback/config.yaml (region + seniority).