Groundwork / Practical playbooks

Good systems begin
with good questions.

Choose one question to work through with your team. Each guide helps you make a useful change to everyday work.

Find your next step

Start where the
work needs you.

No sequence is compulsory. Resolve the important gaps for your chosen workflow.

01 / Start here

Choose work worth improving

Start with a useful outcome and a workflow small enough to understand.

A focused team conversation
Work through this playbook
01

Name the outcome

Write what should improve for a customer or the business. Fewer errors, a faster response, better quality or room for a new service are more useful than “use more AI”.

02

Walk through real work

Ask the people doing it to show a recent example from beginning to end. Record handoffs, waiting, rework and exceptions. Include informal steps that never appear in a process document.

03

Establish a baseline

Choose a representative sample. Record elapsed time, human effort and quality with the same definitions you will use in the pilot. Mark what you have not measured.

04

Agree a decision

Name the owner, scope and result that would justify continuing. Record reasons to stop. Consider simpler process improvements alongside AI.

02 / Build the foundation

Make your knowledge usable

Capture the context that makes good work possible, in the systems that suit your organisation.

Begin with a two-hour working session
Work through this playbook
01

Ask the experienced people

Ask what excellent work looks like, what consistently goes wrong and what a newcomer takes months to learn. Capture examples and the reasoning behind difficult decisions.

02

Find the sources of truth

List the sources a workflow needs. Name an owner, review date and access rules for each. Retain trusted systems of record. A file, database or authenticated business tool can all hold usable knowledge.

03

Separate facts from working context

Distinguish approved standards from current project state and provisional observations. Record where a statement came from. Conflicting or stale sources should remain visible.

04

Test whether it can be used

Try representative tasks. Check whether the person or agent finds the right material, respects access limits and identifies uncertainty. Review additions before treating them as authoritative.

03 / Build the foundation

Define good before you scale

Make quality visible with examples, criteria and evidence from real operation.

Before the first pilot, then at each change
Work through this playbook
01

Write the criteria

Define what a reviewer should check: accuracy, completeness, policy boundaries, useful format and suitability for the intended recipient. Avoid relying on a vague “looks good”.

02

Include the awkward cases

Gather ordinary examples, ambiguous requests, missing information, conflicting sources and costly mistakes. Keep some cases aside to check later changes without repeatedly optimising to the same examples.

03

Compare with the current workflow

Use the same criteria for the existing process and the proposed one. Record review time, correction effort and costs. A faster draft can still create more work downstream.

04

Make the next decision explicit

Agree a quality threshold and errors that require stopping. Review failures by cause: context, model, tool, permissions or workflow design. Record changes and recheck relevant cases.

04 / Design the work

Set authority and accountability

Give a capability enough access to be useful and clear limits on what it may decide.

Before access or authority is granted
Work through this playbook
01

Name the accountable person

Assign an owner who understands the outcome and can change or stop the workflow. Make operational support and escalation responsibilities clear.

02

Describe the boundaries

List the data and tools required. Separate reading, drafting, editing and external commitments. Grant the least access that can complete the agreed task. Keep credentials out of prompts and documents.

03

Define approval and recovery

Specify actions requiring approval, when to ask for clarification and when to stop. Design a usable fallback for failed tools, uncertain outputs and unexpected situations.

04

Review what actually happened

Keep proportionate records of significant actions and decisions. Review incidents and access changes. Check that the person responsible has time and authority to do the job.

05 / Design the work

Design the change with your people

Make the new way of working understandable, useful and open to improvement.

Throughout discovery, design and review
Work through this playbook
01

Involve the people closest to the work

Ask what helps, what gets in the way and what they would protect. Include people receiving the output. Give them a safe way to raise concerns and point out missing context.

02

Make the division of work clear

Show what people will own, what systems will do and where judgement remains essential. Avoid leaving someone accountable for a system they cannot inspect or influence.

03

Make room to learn

Provide examples, practice and clear escalation routes. Budget time for training and review. Discuss where released capacity will go rather than assuming work will reorganise itself.

04

Listen after launch

Review actual use, workarounds, errors and effects on workload. Treat feedback as design evidence. Change incentives or responsibilities when they conflict with the new workflow.

06 / Put it into practice

Run a bounded pilot

Learn whether the proposed design works under realistic conditions before increasing its scope.

A short period with an agreed review date
Work through this playbook
01

Make the starting conditions explicit

Choose a workflow with a named owner, usable context, clear criteria and defined authority. Resolve critical gaps first. Record the baseline and full cost assumptions.

02

Start with limited exposure

Use a safe test environment or reviewed drafts where appropriate. Limit users, task types and permissions. Check failures in integrations and data sources, not just model output.

03

Track the whole job

Measure end-to-end work, including corrections, review, coordination and exceptions. Compare quality and effort with the baseline. Record unexpected effects on customers and staff.

04

Decide together

Review the evidence with the owner and people doing the work. Continue, redesign or stop. Expand only where the observed result and operating arrangements justify it.

Put the pieces together

One workflow.
A better way of working.

Build your company map ↗