Skip to library choices
Richard Smith - Virginia

Decision-first field guide

Get to know how Richard approaches difficult choices.

Understanding how someone works makes trust easier to build.

Decision-first library · Methods for the work

Choose another route

Choose by the decision gap

Match the method to what the decision still needs.

Start with the gap: a disputed metric, unclear options, weak signals, an untested uncertainty, or rationale that may be lost. Each route offers a bounded way to make the next choice more defensible; none makes it for you.

Current playbook · Data-to-Decision Translation

When a metric’s meaning is disputed

Trace its source, owner, and decision use before asking it to carry an argument.

Useful output
A traceable metric-to-decision artifact.
Boundary
It does not turn a metric into a verdict.
Format
Playbook
Trace a metric from source to decision

Current worked example · Decision Modelling

When several interventions are credible

Compare guidance, reviewed assistance, bounded automation, and learning first without hiding their trade-offs.

Useful output
A worked choice, evidence plan, and decision record.
Boundary
It demonstrates a comparison, not a prescription.
Format
Worked example
See four credible choices compared

Current method · Sifting

When a weak signal may matter

Follow friction far enough to distinguish noise, a pattern, or a reason to investigate before committing.

Interactive exercise on the next page.

Useful output
A signal trail and the smallest useful artifact.
Boundary
It supports interpretation, not certainty.
Format
Interactive method
Follow one weak signal through five stages

Current playbook · Operational Intelligence Prototyping

When one uncertainty needs a fair test

Give a prototype an evidence purpose, a focused question, and a condition for stopping.

Useful output
A bounded prototype question and evidence artifact.
Boundary
It is not open-ended discovery.
Format
Playbook
Frame one prototype question and stop rule

Current playbook · Institutional Knowledge Structuring

When rationale may be lost at handover

Connect the choice to its rationale, owner, and review point while the context is still available.

Useful output
A structure for rationale, ownership, and review.
Boundary
It does not automate governance.
Format
Playbook
Structure a decision record that survives handover

Decision-first library · Current questions

AI, people, and accountable decisions

Where can AI help and where must people remain responsible?

Not sure which method fits yet? Start with the situation closest to yours. The examples show where AI may assist—and where context, authority, and review must remain human.

Choose a situation

  • Retrospective + current hypothesis

    Customer context before the model

    Best when: AI is being considered before customers have enough context to choose.

    You will get: questions about access, trust, and whether AI belongs in the experience at all.

    Read why customer context comes before the model

  • Current reflection rooted in earlier work

    A firm workflow boundary

    Best when: interpretation and deterministic enforcement must be separated in a consequential workflow.

    You will get: a boundary model for evidence, validation, and human commitment.

    Trace the VAT-refund decision boundary

  • Historically grounded hypothetical

    Prediction in relation to a human goal

    Best when: a prediction is useful only in relation to a person’s unstated priority.

    You will get: competing traveler needs and trade-offs a system cannot settle alone.

    Compare prediction with passenger priorities

  • Historically grounded hypothetical

    A recommendation that survives live operations

    Best when: recommendations must work inside a busy operation with missing signals and operator authority.

    You will get: reversible triage and fallback questions grounded in arena operations.

    Read how AI would meet live operations where they are

  • Current design reflection

    The handoff to accountable action

    Best when: a probabilistic suggestion must become an accountable action without hiding exceptions.

    You will get: a handoff model for evidence, authority, exceptions, and review.

    Trace the probabilistic-to-deterministic handoff

Decision-first library · Earlier practice, connected

Choose another route

Earlier practice in context

Four foundations that still shape the work.

Choose the foundation that matters to your situation. These themes connect earlier work to the current practice; they are not stages or a required reading order.

Frame the real problem

Practical limits, assumptions, and capacity change what is worth solving.

Earlier sources

Translate the problem together

Business context, technical detail, and competing team priorities must become shared, actionable work.

Earlier sources

Carry the choice into use

A good choice still has to survive UX, backlog detail, launch, adoption, and feedback.

Earlier sources

Build trust around the work

Listening, tailored narratives, follow-through, and patient communication help people understand and support the work.

Earlier sources

Decision-first library · You reached the end

Choose another route

A note for the curious

Thanks for staying with the difficult questions.

Reaching the end of a long field guide takes curiosity and care. Thank you for bringing both.

If one question now feels clearer, this page has done its job. Taking one useful question back to your team is also a complete next step.

Looking for the wider record? The preserved career profile remains available as labelled history. Read the preserved career profile.