Environmental — pollution legal liability (PLL) and contractors pollution liability (CPL) — is a line where the words on the page decide the price. Two accounts with the same revenue and the same requested limit can carry wildly different exposure depending on whether the operations are a clean office park or a fleet of remediation contractors working over aquifers, and whether the sites hold underground storage tanks with a history of releases. That signal is spread across a broker’s submission email and a site pollution application form, in prose, in a site schedule, and in a claims table — and it is almost never summarized on any one line.

Most intake tools read one document at a time and hand back a flat list of fields. That is not how an environmental account is underwritten. InsightXtract runs an agentic, multi-document pipeline that classifies each file, extracts to a per-document schema, consolidates everything into one unified account record by declared source priority, and validates it against your rules and reference data — so the underwriter opens a single, coded, cited record instead of a stack of PDFs.

flowchart LR A[Broker email
+ site pollution
application form
site schedule · claims] --> B[Classify
each document] B --> C[Extract to the
right schema] C --> D[Consolidate
by source priority] D --> E[Link into an
entity graph] E --> F[Validate
rules + reference data] F --> G[One coded,
cited account record]

One submission, two documents, one connected record — classify, extract, consolidate, link, validate.

The idea that governs everything: the exposure is at the site

Before the categories, the single most important concept. In environmental liability, the account-level numbers rarely tell you the risk — the sites do. A company with $40M in revenue and three sites is a very different exposure if one of those sites is a former dry cleaner with underground tanks and an open remediation than if all three are leased office suites. So InsightXtract does not stop at header fields; it reads the site schedule as a table — one row per site, with its own operations, storage-tank count, and remediation history — and the claims history as a table, one row per claim. In the tables below, the Captured column flags how each parameter is held:

  • Current — captured once for this term.
  • Point-in-time — a single static fact about the account.
  • Per-row — a table: one site, or one claim, per row.

A · Insured and identity

Who the account is — the legal party, its identifiers, and its classification. This is the account key that drives clearance and de-dupe. Sources: application form, broker email.

ParameterCapturedWhy it matters
Named insured · DBA / trade namePoint-in-timeThe contract party and unique account key — drives clearance, conflicts, and de-dupe.
FEIN · entity type (legal form)Point-in-timeConfirms identity and legal structure for the policy and aggregation across a group.
Mailing / physical address · primary risk statePoint-in-timeLocates the account and its governing jurisdiction — environmental statutes and cleanup regimes are state-specific.
NAICS code · SIC codePoint-in-timeClass-based rating and appetite / knockout — validated against reference lookups so the code is real, not a typo.

B · Broker and the submission

Who is placing the account and what, exactly, is being asked for — the distribution chain and the top-line terms of the request. Source: broker submission email.

ParameterCapturedWhy it matters
Broker name · broker emailPoint-in-timeDistribution routing, binding authority, and correspondence back to the placing broker.
ProductCurrentThe line being requested — sets the coverage form and workflow.
Effective dateCurrentTerm start and the binding deadline the desk is working against.
Requested limitCurrentThe capacity the broker is asking for — the headline of the ask, reconciled against the application’s stated limits.

C · Operations and exposure base

What the insured actually does — the narrative and the numbers that describe the size and nature of the environmental exposure. Source: application form.

ParameterCapturedWhy it matters
Business description · operations descriptionCurrentThe hazard picture in prose — the difference between a benign operation and one handling contaminants, read and structured, not skimmed.
Operations type (environmental)CurrentThe environmental character of the work — the core appetite and rating signal for this line.
Number of sitesCurrentThe count of physical exposure points — and a cross-check against the site schedule that follows.
Projects per yearCurrentJob frequency for contractors’ pollution — more projects means more chances for a release.
Years in businessPoint-in-timeStability and experience credit — and the depth of the tail an environmental policy may reach back into.
Estimated revenue · total payroll · employee countCurrentThe exposure base for rating — formatted to currency and reconciled across the packet.

D · Coverage requested and limits

The exact shape of the cover being requested — premises pollution vs. contractors, the limit structure, the retention, and the trigger. This is where the ask becomes precise. Source: application form.

ParameterCapturedWhy it matters
Coverage requested (PLL / CPL)CurrentPollution legal liability vs. contractors pollution liability — different exposures, different forms.
Coverage type (premises vs. contractors)CurrentFixed-site risk vs. work performed at others’ sites — drives which exposures are in play.
Each-incident limitCurrentThe per-event capacity offered — the core of the ask, formatted to currency.
Aggregate limitCurrentThe total capacity across the term — the ceiling on the insurer’s exposure.
Retention (SIR / deductible)CurrentThe insured’s share of each loss — drives net exposure and the collectibility question.
Retroactive dateCurrentHow far back claims-made coverage reaches — the single biggest driver of tail exposure on environmental policies.
PremiumCurrentThe target / indicated premium — the price-to-beat and rate baseline, formatted to currency.

E · The site schedule — one row per site

The heart of an environmental submission. InsightXtract reads the schedule of insured sites as a table — one row each — because the exposure is at the site, not the account. Source: application form.

ParameterCapturedWhy it matters
Per site: site addressPer-rowLocates each exposure point — venue, aquifer sensitivity, and proximity to receptors all start with the address.
Per site: operations at the sitePer-rowWhat actually happens on that ground — the site-level hazard that account-level descriptions hide.
Per site: storage tanks (count)Per-rowUnderground and aboveground tanks are the classic release source — a direct frequency-and-severity signal.
Per site: remediation historyPer-rowPrior or ongoing cleanup — the strongest predictor of future environmental cost, captured verbatim per site.

F · Prior insurance

The continuity context — who was on the risk before, which matters more on claims-made environmental cover than almost any other line. Source: application form.

ParameterCapturedWhy it matters
Prior carrierCurrentCoverage continuity and the retroactive-date question — a gap in claims-made cover is a genuine knockout on environmental risk.

G · Claims history — one row per claim

The single biggest pricing input. InsightXtract reads each claim as a row and captures the cost breakdown that shows development — paid, reserve, and incurred. Source: application form.

ParameterCapturedWhy it matters
Per claim: claim number · date of lossPer-rowEach loss’s identity and timing — anchors it to a term and a retroactive date.
Per claim: description · status (open / closed)Per-rowThe nature of the loss and whether exposure is still live on the books.
Per claim: paid · reserve · incurredPer-rowCost, development, and open exposure — a large reserve on an open pollution claim is exactly the signal a snapshot buries.

From two documents to one connected record

Pulling these parameters out of an email and an application is only half the job. The value is in consolidation: the insured named on the application, the broker on the email, the limits in the coverage section, the sites in the schedule, and the losses in the claims table all describe one account. InsightXtract merges them into a single record by a declared source-of-truth priority (the application form wins over the email on coverage and operations details), and links the result into an entity graph — the insured at the centre, connected to broker, submission, coverage, sites, and claims history.

The InsightXtract agent view — an environmental / site pollution submission consolidated into one connected record: the insured linked to broker, submission, coverage, the site schedule, and claims history
The consolidated submission as one coded record — every document’s values linked, each field cited back to its source and validated against reference data.

Why the graph, not just a form

An underwriter doesn’t think in flat fields — they think in relationships: does the number of sites on the application match the rows in the site schedule? Does a site with underground tanks and a remediation history line up with an open claim in the loss table? The entity graph makes those connections explicit and clickable, with every value cited back to the page it came from.

Why it matters to the business

Comprehensive, structured, site-level extraction isn’t a data-entry nicety — it changes the economics and quality of an environmental book:

  • Site and tank exposure surfaced by default. The count of sites, the operations on each, and the number of storage tanks are captured per row — so the underground-tank risk that drives environmental losses reaches the underwriter instead of hiding in a schedule.
  • Gradual vs. sudden pollution priced correctly. The coverage type, the retroactive date, and the operations description together tell you whether you are covering a sudden-and-accidental release or a slow, gradual seepage — the distinction that separates a manageable claim from a decades-long cost.
  • Remediation history read, not skimmed. Prior or ongoing cleanup at a site is the strongest predictor of future environmental cost. Capturing it verbatim per site means known contamination is never underwritten by accident.
  • Venue and jurisdiction made explicit. The risk state and each site address expose the governing cleanup regime and social-inflation venue — environmental statutes and joint-and-several liability vary sharply by state.
  • Consistency and auditability. The same categories, coded the same way, every time — validated against reference lookups and cited to the source document. That is the difference between a repeatable book and one that depends on which underwriter opened the file.

The Environmental agent extracts all of this today — the insured and identity; the broker and submission; the operations and exposure base; the exact PLL/CPL coverage, limits, retention, and retroactive date; prior insurance; the per-site schedule with storage tanks and remediation history; and the per-claim claims history — consolidated into one coded record, every value cited to its source and validated against state and NAICS reference data. The point this post makes is why it matters: the site-level detail and the retroactive date are exactly what move an environmental price, so they’re captured by default. And because it’s all configuration — fields and tables in the agent’s output contract, not code — the schema keeps pace with what underwriters ask for.