AI Work Brief · Free working template
From faster tasks to a better workflow
A worksheet for one recurring piece of work, to decide what is worth testing before anything is bought or built. Read it here, print it or download it to fill in. No email address needed.
Who should use it
- The person accountable for a recurring piece of work, such as an operations lead, a team manager or an owner.
- Anyone deciding whether AI is worth testing on that work before buying a tool or commissioning a build.
- A team that already uses AI and wants to know whether the work itself has improved, not only the first draft.
How to use it
Use this for one recurring piece of work. The purpose is to decide what is worth testing, not to produce an AI maturity score. Blank fields are questions to answer, not evidence that a problem exists.
Version 1.0, 30 September 2026. It is a working template: it contains no findings, savings or client examples.
AI Work Brief · Version 1.0
The worksheet
Six steps, one workflow. Blank fields are questions to answer.
1Name the work
- Workflow
- Accountable owner
- What starts the work?
- What counts as a finished, accepted result?
- Typical volume and measurement period
- Who receives the result?
2Establish the starting point
Record the source and confidence of each estimate. Include the whole process, not only the first draft.
Effort and time for each part of the work, with the evidence behind each estimate Part of the work Human effort per case Waiting or elapsed time Evidence/source Preparation and finding information Producing the initial result Checking and approval Corrections and exceptions Hand-offs and administration - Current quality or acceptance measure
- Important errors and their consequences
- Current tools and incremental costs
3Specify a change worth testing
- What would change in the process?
- What should AI prepare, retrieve, check or automate?
- What must a person judge, approve or own?
- What data, permission or integration is needed?
- What should happen when the system is uncertain or fails?
- Who needs to learn a different way of working?
Compare a simpler process, an existing tool and new engineering where relevant. AI is not a required answer.
4Define a fair test
- Comparable tasks or cases
- Baseline observation period
- Trial observation period
- Quality/acceptance criteria agreed before the test
- Who reviews the output?
- Setup and training effort
- Recurring human effort and operating cost
- What evidence would make us continue, change course or stop?
5Connect improvement to a business result
A reduction in effort is released capacity. It is not automatically cash, revenue or profit.
- Where would released capacity go?
- What specific cost could genuinely be avoided?
- What additional work is there demand to deliver?
- Could the benefit instead be turnaround, quality or reduced workload?
- Which benefit can be measured without counting the same improvement twice?
6Make the next decision
Decision: investigate / run a bounded test / change the process / use an existing product / build / stop.
- Owner
- Next action
- Date
- Evidence still missing
This template is general working guidance, not a prediction of savings or a client case study. Do not enter confidential client or personal data into an unapproved AI tool.
When you want help with it
TechGuidr helps businesses redesign work, make better technology decisions, build what matters and fix what is not working. An AI Work Review can help establish a practical starting point; a scoped AI Work Sprint puts selected changes into practice. Service details and current prices: techguidr.com.
Want a second pair of eyes on the first workflow?
The AI Work Review works through how the work happens today and ends with the workflows worth changing first.