Find where signed contracts deviate from the contract playbook

For: General counsel or legal operations lead who inherited a contract portfolio

Pattern: Map, verify, reduceRuns todayDesigned for 4 to 250 agents

The pain today

Nobody knows which signed agreements carry uncapped liability, odd governing law or weak termination rights. The playbook says what should have been agreed; finding what was agreed means reading every contract.

The ask

I attached our contract playbook with standard and fallback positions, and the signed customer agreements. For each agreement, tell me where liability, indemnity, termination, governing law and data protection clauses differ from our standard, quote the clause, and say whether it is within our fallback or outside it.

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

  • Contract playbook with standard and fallback positions
  • Signed agreements with order forms and amendments
  • List of clause topics to check

The unit of work

One worker task per one clause topic in one agreement.

Why a swarm fits

Each check needs one clause and one playbook position, a very small context. Agreements are independent, and the same topics repeat across the whole portfolio.

Not for

A single contract under ten pages, or a live negotiation markup: one strong model beside the lawyer is cheaper and as good.

The decision tree

5 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

    Is this document a signed agreement, order form or amendment with the customer, rather than a draft, a proposal or an order acknowledgement?

    Sees only: The document's title block, parties and signature page

    Why: Drafts in the folder would otherwise be reported as agreed positions.

    • Yes: 0.60 or higherthenAccept
    • Unsure: 0.30 up to 0.60thenAccept
    • No: below 0.30thenSkip this unit
  2. Before workers, before a task runs

    Small worker or strong modelYes or no, with a probability

    Before a task runs

    Is the clause a short standalone provision, without carve-outs, cross-references or defined terms that change its reach?

    Sees only: The located clause text

    Why: Layered liability clauses need a stronger reader; plain governing-law clauses do not.

    • Yes: 0.60 or higherthenAccept
    • Unsure: 0.40 up to 0.60thenEscalate to a strong model
    • No: below 0.40thenEscalate to a strong model
  3. After workers, the judge checks

    Evidence checkYes or no, with a probability

    After a worker answers

    Does the quoted clause contain the feature the worker relied on to call it outside fallback, such as no cap, a higher cap or a forbidden carve-out?

    Sees only: The quoted clause, its resolved definitions and the playbook position for the topic

    Why: A wrong 'outside policy' label wastes lawyer time; a wrong 'standard' hides exposure.

    • Yes: 0.85 or higherthenAccept
    • Unsure: 0.50 up to 0.85thenEscalate to a strong model
    • No: below 0.50thenReject and retry
  4. Reconciler, while merging

    Conflict checkA choice among options

    While reconciling

    Does the order form or amendment change the clause topic that the main agreement already covers?

    Sees only: The main clause and the order form or amendment passage on the same topic

    Why: Overrides in order forms are where the negotiated exceptions hide.

    • Expressly replaces the main clausethenAccept
    • Touches the topic, silent on which prevailsthenMark unresolved
    • Different topicthenAccept
  5. Accountable person, before anything is settled

    Person decidesYes or no, with a probability

    Before anything is reported as settled

    Is this clause outside the fallback position, so that a lawyer must decide whether to renegotiate, accept or escalate it?

    Sees only: The classified clause and the playbook position

    Why: Classification is not legal advice; the lawyer decides what matters.

    Accountable: A lawyer decides which deviations matter and whether to renegotiate, accept or escalate; classification is not legal advice.

    • 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 turns the playbook into a testable position per topic: standard, acceptable fallback, outside policy.

    Decisions here:1. Scope check

  2. Workers

    Small fast workers from an open-weight family find the clause for one topic in one agreement and classify it against the position.

    Designed for 4 to 250 agents, one worker task per one clause topic in one agreement. 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 checks that the quoted clause supports the classification given.

    Decisions here:3. Evidence check

  4. Reconciler

    A strong reasoning model builds the portfolio view and keeps agreements where amendments change the answer as open items.

    Decisions here:4. Conflict check

  5. Accountable person

    A lawyer decides which deviations matter and whether to renegotiate, accept or escalate; classification is not legal advice.

    Decisions here:5. Person decides

Checked before anything is accepted

  • Every classification quotes the clause and its section reference
  • Order forms and amendments are checked for overrides of the main terms
  • A topic not found is reported as absent, which is itself a deviation for some topics
  • Defined terms used in the clause are resolved before classification

What comes back

  • Deviation table per agreement and topic with quoted clauses
  • Agreements outside fallback, grouped by topic
  • Topics silent in an agreement
  • Clauses where the amendment and main terms conflict

What to measure

  • Classifications a lawyer overturns in a reviewed sample
  • Outside-policy clauses previously unknown to the legal team
  • Lawyer hours per agreement
  • Cost per checked clause against a single-model baseline

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

Settle assignment and change-of-control readings before a transfer

For: In-house or deal counsel planning an intra-group reorganisation, carve-out or sale

Whether consent is needed turns on wording: assignment versus transfer by operation of law, direct versus indirect control, affiliate carve-outs.

Pattern: Cross-examinationNeeds live models5 decisionsDesigned for 6 to 200 agents

Draft and rank fallback wording for contested clauses

For: Commercial counsel refreshing a template or stuck on a contested set of clauses

Counsel drafts one or two alternatives under time pressure and sends the first that reads well.

Pattern: TournamentNeeds live models6 decisionsDesigned for 8 to 150 agents