Review maintenance and test records across a substation fleet

For: Asset manager or maintenance engineer at a transmission or distribution network operator

Pattern: Hierarchical decompositionNeeds a connectorDesigned for 40 to 600 agents

The pain today

Oil analysis reports, breaker timing tests, relay test sheets and battery discharge tests sit inside work orders. Overdue tasks, results drifting toward limits and findings closed without action are found only when one site is audited.

The ask

Go through the maintenance work orders and test reports for our substations. For each transformer, breaker, protection relay and battery bank, tell me which required tasks are overdue under our maintenance policy, which test results were outside the stated limits, and which findings were closed without a recorded action.

Plain words, as you would say it to a colleague. Edit it to fit your case before you send it.

What you attach or connect

  • Work order history from the maintenance system
  • Test and inspection reports as text
  • Maintenance policy with task intervals and limits
  • Asset register by substation and bay

The unit of work

One worker task per one asset's record history.

Why a swarm fits

The fleet is a tree: region, substation, bay, asset. A leaf needs one asset's records and the policy rows for its type. Leaves are independent and roll up into site and fleet views.

Not for

Live condition monitoring or alarm handling. It reads recorded results and does not replace diagnostic judgement on a transformer or breaker.

The decision tree

6 typed decisions, each with an action for every answer

At fixed moments in a run, the engine puts one narrow question to a decision model. The decision model never writes text: it answers yes or no with a probability, picks from listed options, or gives a score, about a small slice of the material. The engine then does exactly what this tree says, which is what makes the run auditable. The thresholds are the template's design values, not measured results.

  1. Planner, while planning

    Split or keep serialYes or no, with a probability

    While planning

    Does the asset register list enough separate bays and asset types at this substation for its records to be read independently?

    Sees only: One substation's rows in the asset register

    Why: Keeps small switching stations as one task and splits only the large sites.

    • Yes: 0.60 or higherthenSplit the work
    • Unsure: 0.40 up to 0.60thenKeep serial
    • No: below 0.40thenKeep serial
  2. Split or keep serialYes or no, with a probability

    While planning

    Do the work orders for this bay name individual assets, so they can be divided per transformer, breaker, relay and battery?

    Sees only: Work order headers for one bay

    Why: Splits to asset level only when records can be assigned without guessing.

    • Yes: 0.60 or higherthenSplit the work
    • Unsure: 0.40 up to 0.60thenKeep serial
    • No: below 0.40thenKeep serial
  3. Scope checkYes or no, with a probability

    Before work starts on a unit

    Does the asset tag or serial on this test report match the asset in the register row?

    Sees only: One test report header and one register row

    Why: Stops a result from being credited to the wrong breaker or transformer.

    • Yes: 0.60 or higherthenAccept
    • Unsure: 0.30 up to 0.60thenEscalate to a strong model
    • No: below 0.30thenSkip this unit
  4. After workers, the judge checks

    Evidence checkA choice among options

    After a worker answers

    Does the quoted test report state a result for this test, and does the supplied policy row give a limit for it?

    Sees only: One quoted test result and the policy row for that asset type

    Why: Uses only the operator's own limits and treats a missing result as missing, never as a pass.

    • Result stated, within the policy limitthenAccept
    • Result stated, outside the policy limitthenMark unresolved
    • Result stated, policy gives no limitthenAsk a person
    • No result statedthenMark unresolved
  5. Evidence checkYes or no, with a probability

    After a worker answers

    Does the quoted work order closure text describe an action taken, rather than closing the finding without one?

    Sees only: One finding and the closure text of its work order

    Why: Finds findings that were closed on paper with nothing recorded as done.

    • Yes: 0.85 or higherthenAccept
    • Unsure: 0.50 up to 0.85thenEscalate to a strong model
    • No: below 0.50thenReject and retry
  6. Accountable person, before anything is settled

    Person decidesYes or no, with a probability

    Before anything is reported as settled

    Does the finding report an out-of-limit result, an overdue task on protection or batteries, or a finding closed without action?

    Sees only: One asset finding with its quotes

    Why: The responsible asset engineer sees every condition-relevant finding; the swarm assesses no equipment.

    Accountable: The responsible asset engineer owns condition assessment and any decision to restrict, repair or replace equipment.

    • Yes: 0.40 or higherthenAsk a person
    • Unsure: 0.15 up to 0.40thenAsk a person
    • No: below 0.15thenAccept

The fleet: who does what

Model tiers by role, not brands: you choose the models. Strong reasoning models plan and reconcile, small fast models do the wide work, and the judge is a decision model from a different family, so it does not share the workers' blind spots.

  1. Planner

    A strong reasoning model builds the tree from the asset register and attaches the policy rows that apply to each asset type.

    Decisions here:1. Split or keep serial2. Split or keep serial3. Scope check

  2. Workers

    Small fast workers from an open-weight family read one asset's work orders and reports against its policy rows.

    Designed for 40 to 600 agents, one worker task per one asset's record history. Each worker receives only its own unit.

  3. Judge, from a different model family

    A decision model from a different family checks each overdue or out-of-limit claim against the quoted record and limit.

    Decisions here:4. Evidence check5. Evidence check

  4. Reconciler

    A strong reasoning model rolls findings up by bay, substation and asset family and keeps records that contradict each other.

  5. Accountable person

    The responsible asset engineer owns condition assessment and any decision to restrict, repair or replace equipment.

    Decisions here:6. Person decides

Checked before anything is accepted

  • Every finding quotes the work order or report and the policy limit applied
  • Missing records are reported as missing, never read as a pass
  • Limits are taken from the supplied policy only, not from general knowledge

What comes back

  • Findings per asset with quotes and dates
  • Overdue task list by substation
  • Asset families with repeated out-of-limit results
  • Records that conflict or could not be matched to an asset

What to measure

  • Findings confirmed by the maintenance engineer
  • Audit findings on sites the swarm had reviewed
  • Engineer hours per substation reviewed
  • Cost per asset

Names of measures only. No result is claimed for this template.

Templates open in the workspace chat with the ask filled in. Nothing runs until you send it.

Get early accessSign in to use

Reconcile study assumptions across an interconnection queue

For: Interconnection manager at a developer, or a transmission planner reviewing clustered studies

Each project's study rests on a base case, on earlier-queued projects assumed in service, on short-circuit levels and on shared network upgrades.

Pattern: Map, verify, reduceNeeds scale6 decisionsDesigned for 30 to 400 agents

Check a flexibility contract portfolio against dispatch obligations

For: Portfolio manager at a flexibility aggregator, or a utility demand-response programme lead

Every site contract has its own notice time, duration limit, activation cap, availability window, baseline method and penalty.

Pattern: Specialist panelNeeds live models6 decisionsDesigned for 8 to 300 agents

Trace grid code and tariff changes to every affected asset

For: Grid compliance or regulatory affairs manager at a generator, storage operator or network company

A revised grid code clause or tariff schedule touches some assets and not others, depending on connection date, voltage level, technology and derogations.

Pattern: WatchtowerNeeds a connector6 decisionsDesigned for 20 to 500 agents