Eighteen decisions for changing an operation with control.

Use the group that matches the decision in front of you. Each guide provides a direct answer, operating conditions, concrete failure modes, a comparison table, and a checklist for the working session.

1. Choose the work

   01

   Workflow automation guide

   Choose a bounded first workflow, then map its states, authority, records, exceptions, and recovery path.

   Read the guide

2. Choose the system

   02

   Build, buy, or integrate an AI system

   Choose configured product, custom software, integration, or a deliberate combination from the operating boundary and exit path.

   Read the guide
3. 03

   Choose cloud, on-premises, or open-source AI

   Compare deployment and operating choices by data boundary, control, latency, continuity, capability, portability, staffing, and total burden.

   Read the guide

4. Control the change

   04

   Run parallel operation before migration

   Run a bounded new path beside a known fallback, compare end states, and expand authority only when stated evidence holds.

   Read the guide
5. 05

   Set operational acceptance criteria

   Turn a promising demonstration into observable conditions for authority, recovery, and continued use.

   Read the guide

6. Own the operation

   06

   Design authority and exception paths

   Decide what a system may complete, what a person must decide, and how uncertain work reaches the right owner.

   Read the guide
7. 07

   Handle uncertainty in retrieval systems

   Separate retrieval, answer generation, and authority so weak, conflicting, stale, or inaccessible evidence receives a controlled response.

   Read the guide
8. 08

   Assign system stewardship before launch

   Give monitoring, changes, incidents, dependencies, documentation, and recovery an accountable home.

   Read the guide

9. Buy the change

   09

   Calculate workflow automation cost and ROI

   Compare the current operating baseline with implementation, transition, and ongoing costs without treating estimates as realized savings.

   Read the guide
10. 10

    Choose an AI implementation partner

    Evaluate whether a partner can carry workflow design, integration, evidence, authority, recovery, and stewardship into real operations.

    Read the guide
11. 11

    Move an AI pilot into production

    Turn a promising pilot into an owned operating system with representative evidence, integration, controls, rollback, and support.

    Read the guide

12. Automate recurring work

    12

    Business process automation examples

    See how recurring intake, quoting, invoice, reporting, and coordination work can move through bounded automated states.

    Read the guide
13. 13

    Document workflow automation

    Receive, preserve, classify, extract, validate, route, and reconcile documents without hiding uncertain or duplicate work.

    Read the guide
14. 14

    Invoice processing automation

    Capture invoices, check vendors and duplicates, route approvals, prepare accounting records, and reconcile exceptions.

    Read the guide
15. 15

    Reporting automation

    Collect named sources, validate freshness and completeness, reconcile totals, route review, and distribute approved reports.

    Read the guide
16. 16

    Data entry automation

    Move information from approved sources into systems of record with validation, duplicate controls, review, and reconciliation.

    Read the guide
17. 17

    Email workflow automation

    Turn approved inbound email into traceable work while preserving message context, review, commitments, and duplicate controls.

    Read the guide
18. 18

    Scheduling automation

    Collect constraints, propose allowed times, confirm commitments, and route conflicts, changes, and exceptions to an owner.

    Read the guide

Start with the choice that is blocking responsible progress: which work to change, which system boundary to use, how to transfer authority, or who will own the operation. Each group keeps the guide beside the decision it answers.

Use the answer, table, failure modes, and checklist in a working session. Record the unresolved conditions before selecting technology or expanding system authority.