Check country handbooks against group people policy

For: Head of people policy or HR compliance lead in a company with staff in many countries

Pattern: Map, verify, reduceNeeds scaleDesigned for 40 to 600 agents

The pain today

Each country has its own handbook and local policies, written at different times. Nobody can say where they contradict the group policy or each other until an employee dispute or an audit finds it.

The ask

I attached our group people policies and the employee handbooks and local policies for every country. For each group rule, tell me what each country document says, where it is stricter, weaker, silent or contradicts the group rule, and show me the exact passages.

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

  • Group people policies
  • Country employee handbooks
  • Local policies and addenda
  • List of group rules to check, if you have one

The unit of work

One worker task per one group rule in one country document.

Why a swarm fits

Each check needs one group rule and the matching sections of one handbook, nothing else. Rules and countries are independent, so the grid of checks runs side by side and only the reconciler sees the whole picture.

Not for

Deciding whether a local rule is lawful. That needs local employment counsel; the swarm only shows what the documents say.

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

    Scope checkYes or no, with a probability

    Before work starts on a unit

    Does this handbook section deal with the subject of the group rule it has been paired with?

    Sees only: One group rule and the heading and opening lines of one handbook section

    Why: Keeps workers off sections that cannot contain the rule, so reading effort goes where an answer can be.

    • Yes: 0.60 or higherthenAccept
    • Unsure: 0.30 up to 0.60thenAccept
    • No: below 0.30thenSkip this unit
  2. After workers, the judge checks

    Evidence checkYes or no, with a probability

    After a worker answers

    Does the quoted handbook passage set a rule that stands to the group rule as the worker reported: stricter, weaker, the same or contradicting?

    Sees only: One group rule, the quoted handbook passage and the worker's grading

    Why: Stops a misgraded cell from entering the rule by country grid.

    • Yes: 0.85 or higherthenAccept
    • Unsure: 0.50 up to 0.85thenEscalate to a strong model
    • No: below 0.50thenReject and retry
  3. Evidence checkYes or no, with a probability

    After a worker answers

    Given every section heading and the matched extracts, does this country handbook contain no passage that addresses the group rule?

    Sees only: One group rule, the handbook's section headings and the extracts searched

    Why: A false silent finding hides a rule that exists, so silence is checked as hard as a quote.

    • Yes: 0.85 or higherthenAccept
    • Unsure: 0.50 up to 0.85thenMark unresolved
    • No: below 0.50thenReject and retry
  4. Reconciler, while merging

    Conflict checkYes or no, with a probability

    While reconciling

    Do these two passages from the same country's handbook and local policy give different answers on the same group rule?

    Sees only: Two quoted passages from one country and the group rule they were matched to

    Why: Keeps internal contradictions visible instead of letting the reconciler pick one.

    • Yes: 0.70 or higherthenMark unresolved
    • Unsure: 0.30 up to 0.70thenEscalate to a strong model
    • No: below 0.30thenAccept
  5. Accountable person, before anything is settled

    Person decidesYes or no, with a probability

    Before anything is reported as settled

    Could this finding lead to a change in what employees in this country are entitled to or required to do?

    Sees only: One grid cell with its quotes and grading

    Why: Anything touching entitlements goes to the policy owner and local counsel before it is reported as settled.

    Accountable: The policy owner and local employment counsel decide which version is correct and what gets rewritten.

    • Yes: 0.30 or higherthenAsk a person
    • Unsure: 0.10 up to 0.30thenAsk a person
    • No: below 0.10thenAccept

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 turns the group policy into checkable rules and maps handbook sections to them.

    Decisions here:1. Scope check

  2. Workers

    Small fast workers from an open-weight family, one per rule and country, each quoting the passage it relies on.

    Designed for 40 to 600 agents, one worker task per one group rule in one country document. Each worker receives only its own unit.

    Decisions here:2. Small worker or strong model

  3. Judge, from a different model family

    A decision model from a different family asks whether the quoted passage supports the stricter, weaker or silent finding.

    Decisions here:3. Evidence check4. Evidence check

  4. Reconciler

    A strong reasoning model builds the rule by country grid and keeps unresolved contradictions open.

    Decisions here:5. Conflict check

  5. Accountable person

    The policy owner and local employment counsel decide which version is correct and what gets rewritten.

    Decisions here:6. Person decides

Checked before anything is accepted

  • Every finding carries a verbatim quote and section reference from the handbook
  • Silent findings are rechecked by a second worker with a wider section scope
  • Translated documents are marked so a native reader confirms the passage
  • The judge rejects findings where the quote does not match the claim

What comes back

  • Rule by country grid: aligned, stricter, weaker, silent or contradicts
  • Quoted passages behind every cell
  • Open contradictions listed for legal review
  • Documents that could not be mapped to any rule

What to measure

  • Share of findings the policy owner confirms on review
  • Reviewer hours per country handbook
  • Contradictions found later that the run missed
  • Cost per checked rule and country pair

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

Check job descriptions against the levelling framework

For: Reward or organisation design lead maintaining a job architecture across functions

Hundreds of job descriptions drift from the levelling framework.

Pattern: Hierarchical decompositionNeeds scale6 decisionsDesigned for 30 to 500 agents

Watch employment law changes against local policies

For: HR compliance lead or employment counsel covering many jurisdictions

Statutory leave, notice, working time and whistleblowing rules change country by country.

Pattern: WatchtowerNeeds a connector5 decisionsDesigned for 20 to 400 agents

Stress-test payroll rules drawn from collective agreements

For: Payroll manager or labour relations lead configuring pay rules from collective agreements

Overtime, shift premiums and allowances are configured from long collective agreements.

Pattern: Adversarial reviewNeeds live models6 decisionsDesigned for 12 to 240 agents