---
name: adopt-freed-time-agreement
description: Prepares a team conversation guide and a one-page record of which tasks now take less effort, what the team stops doing, where freed time goes, what does not expand and a workload check date. Use for "run adopt-freed-time-agreement", "freed time agreement", "what happens to the time AI saves", "AI just means more work", "time saved turns into more tasks", "where does saved time go", "team conversation about workload and AI", "agree what we stop doing", part of the Claude Guide for AI Adoption Pack by Polar Bear.
---
# Freed Time Agreement
## When To Use
The time AI saves quietly turns into more tasks and longer days, and nobody decided that. Use this once a team has used AI on real work for a while and you want the question asked out loud. It answers one question: now that some tasks are lighter, what do we stop, where does the freed time go, and what stays the size it was?

## When Not To Use
If the problem is finding time to learn before the change starts, run Adoption Capacity Check. If you need the programme's measures with definitions and baselines, run AI Value Measures: this is one team's conversation and record, with a single workload measure.

## Inputs
- The team and its main tasks, and which of them AI is now used for (by task, never by person).
- What has been added to the team's work since it started, in the team's words, and any workload signal the team already has (work outside hours, backlog age, rework) and where it comes from.
If you have none of this, I start from the team's task list and the four questions and mark the output as a first draft for the conversation.

## Approach
The record borrows the balancing measure from the Institute for Healthcare Improvement's guidance on establishing measures: a check on whether an improvement in one place creates a problem in another. It reads workload through the demands area of the Management Standards of the UK Health and Safety Executive. As context only: one study of one employer, still in progress, reported people working faster, on a wider scope and for longer hours. The judgment is that lighter tasks need a decision, or the gap fills itself. The failure it prevents: the report that used to fill a morning gets lighter, nobody says anything, and a quarter later the team produces more of everything and works later, without anyone having chosen that.

## Workflow
1. Ask three questions: which team is this and who runs the conversation, which tasks is AI used for, and what workload signal does the team already have?
2. Write the conversation guide around four questions for the team: which tasks take less effort now, what has quietly been added since, what will we stop doing, and where should freed time go?
3. Record each task as "lighter", "about the same" or "heavier", in the team's words. No hour count is invented or estimated. If the input holds time data per person, I work at team level, drop the names and say so.
4. Name what stops and where freed time goes: quality, a backlog, learning, recovery. A stop needs a person who tells whoever used to receive that work; an ending nobody announces comes back.
5. Fill "does not expand" with the scope, hours and expectations the team controls. Headcount, targets and contracts are not decided here: list them as questions for the people who own them. Working time questions: check with a qualified adviser.
6. Pick one balancing measure for workload that the team already has or can observe, at team level, with its source. You set what counts as a warning. Then set the workload check date and who calls it. Close by reading the record back: it says what the team agreed, never what leadership guarantees.

## Output Format
```markdown
# Freed Time Agreement
Team: [team] | Conversation on: [date] | Run by: [manager or facilitator] | Status: [draft for the conversation / agreed by the team]
## Conversation guide
| Question for the team | What to listen for |
|---|---|
| Which tasks take less effort now? | [the team's words, no hour counts] |
| What has quietly been added since? | [new tasks, wider scope, later hours] |
| What will we stop doing? | [the old step, the duplicate check, the report nobody reads] |
| Where should freed time go? | [quality / a backlog / learning / recovery] |

## Record: tasks and effort (the team's own account)
| Task | Lighter, about the same or heavier | What has been added | What stops | Who tells the people affected |
|---|---|---|---|---|
| [task] | [word] | [addition or "nothing"] | [what stops] | [role] |

## Where freed time goes
- [quality / a backlog / learning / recovery]: [what this means for this team]
## What does not expand
| The team controls | Agreed limit |
|---|---|
| [scope / hours / expectations] | [limit in the team's words] |

Questions for the people who own them (not decided here): [headcount / targets / contracts] to [owner].
## Workload check
Balancing measure: [measure the team already has or can observe] | Source: [source] | Warning when: [you set it] | Check date: [date] | Called by: [name]
## Decision
[team] and [manager] agree this record by [date]; [manager] takes the open questions to [owner] by [date]; [name] calls the workload check on [date].
```

## Done When
- Every task carries a word from the team and no figure, and at least one thing stops, with someone who tells the people affected.
- "Does not expand" lists only what the team controls; the rest sits as questions with an owner.
- One balancing measure has a source and a warning you set, and the check date has a name beside it.

## Quality Bar
- The study cited is described with its limits (one employer, in progress) and no figure.
- Nothing about jobs, headcount or contracts is stated or implied; working time questions go to a qualified adviser.
- No promise: Claude prepares the guide and the record; the team has the conversation and the record states what it agreed.
- The record holds the team's own account and no saving figure; nobody's time is measured individually, and workload and headcount decisions stay with the people who own them.

## Next
Run adopt-ai-rules-explainer (AI Rules Explainer) to make the rules people keep asking about readable.

## About the makers
This pack is made by Polar Bear, a consultancy built by ex-McKinsey founders with a dream to make AI work for People, not instead of them. We help our clients build people systems and AI-first ways of working, and we run our own company on Claude. If your team has outgrown the self-serve version, message Pauline (linkedin.com/in/paulinebertry).