Principles
What AI should and shouldn't do — how we think about it.
Timeback shows time-back estimates for 800+ occupations. Numbers out of context can read the wrong way, so this page is the stance behind them — written by Clear Road Labs, the team that builds the systems.
See also: how we work · security & data handling
Time reclaimed ≠ jobs eliminated.
What we model is task-level time — the documentation, coordination, and research work that sits inside a role. Our clients use the reclaimed capacity for the backlog they've never had time for, higher-judgment work, or customers they couldn't serve before. If your goal is reducing headcount, we're not the right partner.
Humans stay in the loop.
Every blueprint we build has explicit checkpoints where a person approves, edits, or rejects the AI's output. We name these in the blueprint and they aren't optional. Agents draft; people decide.
Estimates are ranges, not promises.
The minutes and dollars on this site are median estimates derived from O*NET task descriptions. Real time-back depends on your workflows, your tools, and your team's adoption speed. We refine every number during the scoping engagement before anything gets committed to a statement of work.
Judgment, creativity, and care don't automate.
We score tasks that require physical presence, creative judgment, empathy, or high-stakes accountability lower on purpose. The highest-impact agents we build free people up for exactly that kind of work — they don't try to replace it.
If something reads the wrong way, tell us.
We wrote the copy on this site, and we update it when it stops matching how we work. If a number, a label, or a page reads differently than we intended — especially if it lands badly on someone in the occupation being described — we want to hear it.
damon@clearroadlabs.com