Field guide 01 / Discovery

Map a manual loop before choosing a tool.

A practical way to make one repeated workflow visible before discussing software, automation, or AI.

Why this guide exists

Most system projects begin with a tool name and work backward. This guide begins with one operating loop and records what people, information, decisions, and exceptions actually carry it.

Work through the decision

Four steps, with the unknowns left visible.

  1. 01

    Choose one loop

    Use a repeated piece of work with a recognizable beginning and end. “Run operations” is too broad; “turn an approved order into a scheduled job” is bounded enough to inspect.

    • What event starts the work?
    • What observable state means it is finished?
    • How often does it happen, and what volume is known?
  2. 02

    Observe the work that happens

    Follow a representative example through the people and tools involved. Record the actual path, including side messages and personal workarounds, rather than documenting the policy version.

    • Who touches the work and what do they need to know?
    • Where is each fact first created, copied, or changed?
    • Which decisions, waits, and exceptions alter the path?
  3. 03

    Mark the coordination load

    Identify where a person bridges two systems, asks for status, reconciles versions, or keeps the next step in memory. Leave time and cost unknown unless they were actually measured.

    • Where is the same fact entered again?
    • Where does someone chase, check, or translate?
    • What stops when one person is absent?
  4. 04

    Frame the next decision

    State what the operation needs to make possible and which evidence is still missing. This creates a useful discovery decision without jumping to a product or implementation quote.

    • Which operating change would remove the most uncertainty or coordination?
    • What constraint or exception must remain true?
    • What needs to be measured before making an investment claim?
Completion check

Keep these visible.

  • One specific trigger and one observable end state
  • The roles involved, including who decides and who handles exceptions
  • Inputs, source records, state changes, and handoffs
  • Repeated entry, waiting, chasing, reconciliation, and rework marked
  • Unknowns left visible instead of filled with assumptions
  • The next operating decision stated without prescribing a product
Watch for

Common ways the work drifts.

  • Mapping an entire department instead of one bounded loop
  • Documenting the intended policy instead of observing real work
  • Turning every inconvenience into a software requirement
  • Inventing time, cost, or error estimates because measurement is inconvenient
Use the guide on a real decision

Bring the operating context, not a polished answer.

An Ignite request can begin with the loop, blueprint decision, or ownership gap this guide helped you make visible.