Every decision you make, every project you reopen, every session left waiting: finally counted, finally visible. The work nobody sees, and what it takes from you.
Example, fictional data
Thursday: 12 project switches before noon. 3 evenings out of 5 went past 11 pm.
Scroll the chart sideways to see the whole day.
A preview of the upcoming journal. Each day runs from 5 am to 5 am the next day; each cell is a quarter hour. Top row: quarter hours with you, the quarter hours when you wrote to an agent, in the project’s color. Below: the number of active sessions.
| Day | Quarter hours with you | Projects | Project switches | Time span |
|---|---|---|---|---|
| Monday | 42 | 4 | 19 | 9:45 am–12:45 am |
| Tuesday | 21 | 4 | 7 | 10:15 am–6:30 pm |
| Wednesday | 35 | 3 | 8 | 9:30 am–1 am |
| Thursday | 37 | 5 | 21 | 8:30 am–11:15 pm |
| Friday | 16 | 4 | 5 | 10 am–5:15 pm |
Example, fictional data
151
19.5M
60
3
Your agents produced 3 times as many tokens as the previous week, for 1.6 times as many quarter hours with you.
Counts, compared with your previous week: no score, no predictions. Alongside them, your own end-of-day rating, if you give one.
It looks at how you use AI, not at you. If you allow it, it will open a PR in your repository, visible to anyone with access to it: you’ll merge it or not.
Nothing to fill in: collection will be automatic. The only optional input is your own end-of-day rating, in one keystroke.
A commitment for the upcoming product. Your summary is yours first; you share it only if you choose to.
| What is measured | You | Your team and its manager |
|---|---|---|
| Your load | Yes | Aggregated with others, with your consent |
| Your journal | Yes | No |
| Your retro | Yes, if you turn it on | No, except a PR you allow |
| The text of your conversations | Yes | Never |
14%
of US workers surveyed who use AI say they have experienced “brain fry”. Those most at risk juggle several tools or oversee several agents.
“Developers are having to make decisions more often.”
For the upcoming product.
No. Collection will be automatic. You can add an end-of-day rating in one keystroke, if you want to.
No. A team version will only show aggregates, with your consent. Before any team use, talk to your data protection officer and your employee representatives.
Yes. It will be part of your retro, which you choose to turn on or not. Before launch, we’ll explain where the analysis will run.
Not without you. Your retro will propose its changes in a PR, which only reaches your main branch if you merge it. The first version will only suggest instructions in AGENTS.md or CLAUDE.md: no permissions, no hooks.
A guardrail enforces a rule you’ve already written. Humanload will suggest instructions based on your repeated reminders, then count the reminders that come back.
We don’t give legal advice: your data protection officer and your employee representatives are the right people to ask. Humanload won’t score or rank people. Load will be shown as event counts.
We haven’t announced a launch date yet. We’ll email you when it’s available.
For a team load assessment based on aggregate data, email hello@humanload.app