Clause library and precedent search: the bound wordings that answer the broker, cited, in seconds
A cited precedent in seconds, not an email to the one specialist who knows.
See one case, screen by screen ↓A wordings solution that shows how a clause was worded before, citing the bound wordings. An underwriter asks in plain words — “Have we bound a cargo sanctions clause without the US reference?” — and gets the versions the carrier actually bound, the changed words, who approved or refused them and what happened at claim.
How to choose AI solutions for insurance →
Updated
The broker's draft lands on the morning of the renewal call with “only a small change” to the sanctions clause: the United States reference gone, the trigger back to “would expose”. Whether the carrier has agreed that before — for whom, in which year, with whose approval, and what happened at claim — sits in thousands of bound wordings and in the memory of one or two wordings specialists.
So the question goes by email to the wordings team, and the answer comes back in a day or three while the broker wants it on the call. Without the precedent, an underwriter concedes a term the carrier refused four times, or refuses one it accepted for another insured, and a non-standard clause agreed in a hurry turns into a coverage dispute years later — an outcome that rarely gets back to the underwriter who will be asked again.
Estimated split for one clause question answered by email to the wordings team (1–3 days, from wordings-team practice), against the same question asked in the solution, where the search itself takes seconds.
Six specialist agents find, compare and cite every bound version of a clause; the underwriter decides.
Tellurin Commodities B.V. renews its marine cargo open cover on 1 November: $25,000,000 any one conveyance, premium $1,840,000. The broker’s wording arrives at 09:41 and the call is at 15:00. Here is the day, screen by screen, in the working solution.
Sophie Lang of Marlow Keane Brokers sends a 14-clause renewal wording. Hannah presses “Check all 14 clauses”: the wording reader splits the draft, the difference marker compares each clause with the carrier’s standard, and the precedent finder searches 38,412 bound wordings for the four that changed. Ten are our standard word for word, two were bound before with approval, one was refused before — four times — and one has never been seen in the book.
Clause 7, “MK Sanctions Clause 2026”: “US reference removed; ‘would expose’” — refused before.
She types the question the way she would ask a colleague: “Have we bound a cargo sanctions clause without the US reference?” The run is on screen as it happens — the question read as the sanctions family in marine cargo, 2021–2026; 412 bound wordings and refused proposals searched; nine versions matching, six answering; the changed words marked; approvals, refusals and the one claim attached.
“412 bound wordings and refused proposals · 9 versions match, 6 answer the question.”
The answer is a stack of precedents, each a wording the carrier actually bound, with the insured, the year, the limit, the reference and what happened next. Beside it, the cited answer: “Once, and it went badly.” The US reference came out once, for Corvane Grain in March 2022, and has been refused four times since.
One click fans the six versions side by side. Added words are lit green, removed words struck in red, and a sanctions regime added or removed is marked in amber; unchanged runs fold away so each version shows only what moved. 53 words changed across the six — from Aldersgate Coffee Traders on SC-2019 to Brennholt Agrar on SC-2024 with the carve-back BRC-1.
The difference marker never judges whether a change is acceptable — that is the underwriter’s decision.
She opens the Corvane version against the one before it. Three words changed, and “or United States of America” is struck. The decision line reads: approved by Julian Whitlock, 18 Mar 2022, “single-year exception”. The outcome line: claim C-23-0418, wheat cargo, $1.9m, buyer on a US list only, argued seven months, settled $1.42m.
PSI-MC-22-0391 opens at its sanctions clause, with margin notes from the agents: 24 words differ from SC-2024, the exception approved, and the claim. With the US reference removed the clause did not suspend payment, and the reinsurers’ US-owned panel would not follow. Argued for 214 days; settled at $1,420,000.
Claims counsel’s full note stays with claims — underwriters see the outcome line only.
The precedent table counts the same request across every class and year: 41 wordings bound without the US reference, 42 refused, and none bound in marine cargo since March 2022. In marine cargo, in order: Halvard Shipping refused by Hannah in November 2022, Varga Cereals by Tom Hadley in April 2023, Kestrel Agri in June 2024 and Sundby Timber in January 2026.
Wording drift splits the sanctions clause into phrases across 1,947 bound wordings and 23 versions. “May expose” went from 0% of 2021 wordings to 98% in 2026; the US reference removed peaked at 6% in 2022 and is 0.3% now. A drift alert is already raised: “would expose” is still in two renewals due in November, and their owners were told on 9 October.
The note writer drafts the position from the stack: we will not remove the United States reference. What we offer is SC-2024 with the blocking-regulation carve-back BRC-1, as bound for Ostrava Metals, Lemaire Négoce and Brennholt Agrar — Tellurin is domiciled in Rotterdam, so the carve-back applies. What not to concede, and who to refer to if the broker insists: Julian Whitlock, Head of Wordings, before the call ends.
“Nothing is sent to the broker by an agent.”
Hannah signs: she decides to offer SC-2024 with BRC-1 to Marlow Keane for Tellurin Commodities B.V. and to refuse removal of the US reference. It is within her authority because that version was bound before with the Head of Wordings’ approval. Her signature and the four citations are recorded, and she goes into the call with the precedent in hand.
“Signed · recorded as the underwriting decision.”
The same sanctions clause, seen by the four people who carry it — the underwriter on the call, the Head of Wordings, claims counsel and the underwriting assistant.
Six specialist agents find, compare and cite every bound version of a clause; the underwriter decides.
Reads every bound wording, endorsement and standard form, splits it into clauses, and names each clause’s family and version.
Turns a plain question into clause family, class, years and markets, and ranks the bound versions that answer it.
Compares versions word by word — against the version before, our standard and the broker’s draft — and marks each change as scope, trigger, sanctions regime or cosmetic.
Attaches approvals, referrals, refusals and claim outcomes to the clause version that was used.
Measures each phrase’s share of the book by quarter and raises an alert to the family owner when a non-standard phrase passes the threshold.
Drafts the negotiation note from the precedent stack: the position, the counter-wording quoted exactly as bound, and what not to concede.
Decides; Head of Wordings approves anything never bound. The agents propose; a named person decides.
The assistant answers from bound wordings, standard forms and approval notes only — and when it has no wording to cite, it says so and asks for the clause family or the class.
Draft the counter-wording for Tellurin
Drafted on your note: our SC-2024 with the blocking-regulation carve-back BRC-1, word for word as bound for Ostrava Metals and Brennholt Agrar. Tellurin is domiciled in Rotterdam, so the carve-back applies. It is within your authority because it was bound before with approval.
Show me the claim behind the Corvane wording
Claim C-23-0418: a wheat cargo loss of $1.9m in August 2023. Because the US reference had been removed, coverage was argued for seven months before settling at $1.42m. Claims counsel’s full note stays with claims; this is the outcome line.
Which of my November renewals still say “would expose”?
Two: PSI-EN-25-0377 (energy) and PSI-PR-25-0112 (political risk) still carry the SC-2019 trigger. 96% of 2025 wordings use “may expose”. The owners were told on 9 October.
Tell me if the US reference is removed anywhere again
Watching it. If any bound wording, in any class, drops the US reference from the sanctions clause, I will tell you and Julian Whitlock the same day. The alert is in Settings where the team can see it.
13 screens from the working solution, on its sample data. Pick one to see it large.
Who may accept what, where the precedents come from and what the agents may never do — all settings.
| Setting | Default | Choose from |
|---|---|---|
| Version bound before with approval | Class underwriter | Class underwriter · Senior underwriter · Head of Wordings |
| Version refused before or never bound | Julian Whitlock | Julian Whitlock · Wordings committee |
| Sanctions and cyber families: any change also copies compliance | On | On · Off |
| A negotiation note is a decision | On | On · Off |
| No citation, no sentence | On | On · Off |
| Claims counsel notes: outside claims, show the outcome line only | On | On · Off |
| Drift alert threshold | 5% of a quarter’s new wordings | 2% · 5% · 10% |
| Search the refused proposals too | On | On · Off |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample book of wordings · “target” = the design goal, measured in the live solution · “estimated” = our estimate (1–3 days from wordings-team practice) · “typical” = published figures (LMA wordings page: 50–100 new or updated wordings a year). The carve-back BRC-1 refers to Council Regulation (EC) No 2271/96, as shown in the solution. People, insureds, brokers and wordings named on this page are fictional — characters and sample data in the working solution.
A wordings solution that shows how a clause was worded before, citing the bound wordings. An underwriter asks in plain words — “Have we bound a cargo sanctions clause without the US reference?” — and gets the versions the carrier actually bound, the changed words, who approved or refused them and what happened at claim.
Market libraries and standard-form tools hold the market’s or your standard language. This answers a different question: have we agreed this before, for whom, with whose approval, and what happened at claim. It reads your own bound wordings, endorsements, approval and refusal notes and claim files, alongside your standard forms and the market model clauses.
Yes. It splits the broker’s draft into clauses and checks each one against the bound book. In the working solution a 14-clause renewal wording came back as 10 our standard word for word, 2 bound before with approval, 1 refused before — four times — and 1 never seen, which goes to the Head of Wordings.
Both. It searches the broker versions the carrier declined, with who declined them, so the answer can say a request was accepted once and refused four times since. You can switch searching refused proposals off in Settings.
A view of how one clause family’s language moved across the whole book, year by year: each phrase as a share of that year’s bound wordings, with the events behind the changes. The drift tracker raises an alert when a non-standard phrase passes the threshold you set — 5% of a quarter’s new wordings by default.
No. Outside claims, a claim linked to a clause shows as one outcome line — the loss, the result and the days in dispute. Claims counsel’s full notes stay with claims.
The underwriter decides. A version bound before with approval can be offered within their authority; anything refused before or never bound needs the Head of Wordings. The note writer drafts the negotiation note, but no agent sends anything to a broker, edits a wording or signs.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your standard forms, authority rules and a sample of bound wordings, approval notes and claim outcomes — and it goes live once every quality gate has passed. We will walk you through it on your own wordings first.
Control gaps found before the quote, each linked to the answer or scan.
1-min demoCyber Accumulation ManagementExposed insureds and limit listed the day a vendor event breaks.
1-min demoPolicy Wording ComparisonEvery deviation from your form found in minutes, none bound without an approver.
1-min demoWe’ll run Clause Library & Precedent Search on a sample of your own bound wordings and a broker draft.