# AI automation project brief

An editable template from Agents Autonomous: https://agentsautonomous.ai/how-we-work

Use a general description first. Do not include passwords, customer records, personal data, or confidential documents. This template prepares a conversation; it is not a quote or an agreement.

## 1. One job to improve

- Task and team:
- What starts the work:
- What a completed task produces:
- Who owns the current process:
- What is outside this first scope:

## 2. How the work happens today

- Current steps and handoffs:
- Frequency and approximate volume:
- Time spent doing, checking, correcting, and waiting:
- Where that baseline came from, and the period measured:
- The most common exception:

## 3. Systems and information

- Systems involved and their owners:
- Information that may be read:
- Information that may be changed:
- Who can authorize each permission:
- Known limits on access, retention, sharing, or integrations:
- Whether redacted or synthetic examples are available:

## 4. The smallest useful result

- First users and the journey to support:
- Proposed output (answer, draft, record, report, or action):
- What an existing feature or rules-based workflow could already handle:
- Where interpreting variable information might help:
- What remains a human decision:

## 5. Acceptance

- Representative normal case:
- Missing or conflicting information case:
- Duplicate or repeated request case:
- Permission or unavailable-system case:
- Expected behavior and evidence for each case:
- Quality floor and unacceptable failures:
- Named reviewer and decision date:

Use the companion worksheet: https://agentsautonomous.ai/proof#acceptance-worksheet

## 6. Operating boundaries

- Actions that require approval:
- Conditions that must stop the workflow:
- Manual fallback when a tool or model fails:
- Person responsible for reviewing failures:
- How changes will be tested before release:

## 7. Cost and timing assumptions

- Available budget or range, if known:
- Dependencies on source access and staff availability:
- One-time work versus recurring operating costs:
- Review and correction effort to include in cost:
- Timing constraints and their reason:

These are inputs for scoping, not a delivery promise. Compare operating cost on accepted tasks, not model calls alone: https://agentsautonomous.ai/lp/ai-cost-control#cost-worksheet

## 8. Decision

- Proposed next step: configure existing tools / investigate / prototype / pause
- Evidence still needed:
- Decision owner:

Discuss the outline: https://agentsautonomous.ai/contact