Airclerk helps loss adjusting firms put AI to work on the reports that follow a site visit: the claim form, photos, readings and quotes turned into a draft in the instructing insurer’s template, with what the insured said kept apart from what the adjuster saw. The draft arrives with its gaps listed. Causation and the reserve stay with the adjuster, and cover stays with the insurer.
Illustrative example
A preliminary report on an escape of water at a light-industrial showroom, drafted from the claim form, the site photos, moisture readings, the plumber’s invoice, two builders’ quotes and the electrician’s test report. The draft keeps the insured’s account apart from what the photos show, sets the quotes side by side and leaves cover to the insurer. Illustrative claim; no real insured.
| Item | Insured’s builder | Second quote | Tied to |
|---|---|---|---|
| Ceiling tiles and grid, whole showroom, 64 m² | 6,850 | 5,920 | 18 tiles stained (photos 08 to 21). Both builders price the whole ceiling to match; for the adjuster to confirm. |
| Carpet tiles, 112 m² | 9,400 | 9,180 | Readings F1 to F14 |
| Wall linings, east wall, lower 600 mm | 3,200 | 3,050 | Readings W1 to W8 |
| LED ceiling panels | 4,320 | 1,440 | Twelve panels against four. The test report records four failed. |
| Electrical inspection and test | 480 | 480 | Electrician’s report, 13 October |
| Painting, ceiling and east wall | 2,100 | 1,940 | Both quotes |
| Drying equipment, nine days | 2,430 | 2,430 | Four days invoiced to date, five estimated from the drying log |
| Material damage, before excess | 28,780 | 24,440 |
Every line in the draft says what kind of statement it is: what the insured said, what the photos show, what the plumber wrote, a reasoned guess at cause with its confidence, and a question about cover that is set out rather than answered. The quotes are compared line by line, and the lighting explains 2,880 of the gap between them. The other 1,460 is spread thinly across the rest, and none of it is averaged away. The adjuster opens the draft and does the adjusting.
The division of labour is written into the report. The workflow drafts: it assembles the file, keeps each kind of statement in its place, compares the quotes and lists what is missing. The adjuster decides causation and the reserve, writes the recommendation and signs the report. Cover is the insurer’s decision, and the draft does not pretend otherwise.
The same discipline runs through nine other workflows, from the final report to the risk survey →
Loss adjusters can use AI models such as Claude, Microsoft Copilot or ChatGPT for the reporting that follows a site visit: preliminary, interim and final loss reports drafted in the instructing insurer's template, scopes of damage and quotes compared line by line, the policy wording's relevant clauses set beside the facts, business interruption claims assembled from the insured's records, claimant updates drafted from where the file has reached, and risk survey reports for underwriters. It works from what the firm already keeps, such as the claims system, email, photos and document folders. Claim isolation and sourced figures are built into the workflow, and nothing reaches an insurer or a claimant without the adjuster's approval. The AI drafts the report. It does not decide cover or sign the report.
Each is a candidate first workflow, adapted to the firm’s own templates and claims system, and to its instructing insurers’ reporting guidelines. These are starting patterns rather than a packaged product: each is built against the firm’s own closed files before anyone relies on it. The card marked NZ applies in New Zealand only.
The first report after the site visit, drafted in the instructing insurer's template from the claim form, site notes, photos, readings and the policy schedule. The insured's account, what was seen and what was measured stay in separate places, and the draft names what is still needed before the next report.
Quotes broken into line items and compared against the scope and against each other. Differences are explained from the evidence, items outside the scope are flagged, and nothing is averaged to close a gap. Rates come from the quotes and the firm's own schedules, never from memory.
Each later report drafted from the last one and the file since: what has changed, what has been settled, how the reserve has moved and why, and what is still open. The insurer can follow the claim without rereading the whole file.
The clauses that bear on the claim, taken from the wording and schedule actually on file, with the facts pointing each way set beside them. It drafts considerations for the adjuster. It does not decide cover.
The wording and schedule set the basis of the calculation. The workflow assembles the sales records and accounts behind it, each figure tied to the document it came from and the missing months listed. Adjustments and assumptions are for the adjuster or the forensic accountant to set.
For residential claims that involve the Natural Hazards Commission, the file is organised by date of loss, which decides whether the Natural Hazards Insurance Act 2023 or the old EQC Act applies, and the NHC portion is kept apart from the private insurer's. Apportionment stays with the adjuster.
Status updates for the insured and the instructing insurer, drafted from where the file has actually reached and in line with that insurer's guidelines. The adjuster reads and sends them. Nothing goes out by itself.
After a severe weather event, new instructions sorted by severity and location, with possible vulnerability flagged for a person to follow up. It gives the firm a view of the queue. Who attends first is still a person's call.
Survey notes, photos, plans and fire protection records drafted into a risk survey report for the underwriter: construction, occupancy, protection and exposure, with recommendations ranked and each tied to what was seen. The surveyor sets the ranking and signs the report.
The firm's own reports checked before they go out, against the instructing insurer's reporting guidelines and service levels: sections missing, figures that disagree from one page to the next, photos referred to but not attached, and reports close to missing the insurer's service level.
Optional AI Process Assurance can sit behind the workflows we implement: the instructions, sources, exceptions, review changes and sign-off retained as an AI Workpaper, so the firm can show an instructing insurer what the AI did, what it did not do, how the result was checked and who approved it. It is a platform subscription, implemented alongside the workflow, and the workflow runs without it.
A free introductory conversation is enough to say whether the report you have in mind is a first workflow, and to quote a fixed fee for it. The detailed design happens inside the engagement, with your adjusters, against your own templates and closed files.
Where AI could make the biggest difference, how that work runs today, the systems it touches and who owns it. We also work out together whether Airclerk is the right partner for it. If we are, the conversation gives us enough to quote.
One document: the outcome we are aiming for, what we deliver, what your team contributes, the fixed fee and timeline, and what acceptance looks like. Nothing starts until it is agreed on both sides.
The engagement opens with the detailed design, agreed with you before anything is built. Then we configure and connect the workflow to your documents and systems, build in the agreed review controls and audit trail, test it on real work, and prepare the people who will run it.
Handover, and the support that follows go-live, are written into the scope so you know what is covered before you sign. Ongoing operations and further workflows are a separate agreement, made when you want them rather than switched on by default.
Read how an engagement runs in detail →
Airclerk can implement Claude, Microsoft Copilot or ChatGPT depending on the firm’s existing environment. Workflows run on the firm’s approved enterprise instance and under its own vendor agreement.
The first workflow is chosen by volume, how much of the file arrives in a usable form, repeatability and review effort. We would usually suggest the preliminary report or the quote comparison, where the adjuster already does the thinking and loses hours to assembly. Views on cover, settlement offers and anything sent to a claimant stay outside the first workflow until it has proved itself.
Each draft carries the name of the adjuster who will check and sign it. The workflow stops at that gate. It does not send a report to an insurer or an update to a claimant by itself, and it does not instruct a contractor. Where AI Process Assurance is enabled, the AI Workpaper records which sources were used, what exceptions were raised, what the adjuster changed and whether the gate was passed.
The workflow gets only what the adjuster had when a past report was written: the claim form, photos, readings, quotes and wording. The report the adjuster actually wrote, and how the claim ended, are held back. Senior adjusters then score the drafts for material errors, missed issues, figures without a source, and places where what the insured said has been written up as fact. Reports from the same claim are kept together, so a later report cannot give away the answer to an earlier one.
The engagement produces: the configured workflow on the firm’s approved AI platform, connected to its records; a Workflow Charter defining its stages, permitted sources and approval gate; a boundary statement describing what the workflow must never decide or do; evaluation results from the closed-file testing; a runbook for adjusters and support staff; training for each role; a retained AI Workpaper for each material run where Assurance is enabled; and the check-ins after go-live that confirm the workflow is being used safely, shaped to the firm and written into the scope. Ongoing operations beyond that are a separate agreement.
An adjuster gives a view on cover; the insurer decides it. The draft sets out the clauses and the facts on each side, and stops there. It never tells a claimant what is or is not covered.
What the insured said is recorded as what they said, with where and when they said it. It is not written up as fact because it is plausible, and it is not discounted because it is inconvenient. Where the evidence disagrees with it, the draft puts the two side by side.
Repair rates, quantities, excesses, sums insured and limits come from the quote, the schedule, the wording or the firm’s own rate schedule, each named. Where a figure is missing, the draft names the gap instead of supplying a likely number. A plausible number in a loss report has a way of becoming the reserve.
Photos are filed against the observation they support, and the adjuster checks anything described from them. A model can say what a photo shows. It cannot say how long a stain has been there. An assessment made from photos alone is labelled as a desktop assessment, not a site inspection.
Loss adjusters can use AI for the reporting that follows a site visit: turning the claim form, site notes, photos, readings and quotes into a draft preliminary report in the instructing insurer's template, comparing quotes line by line against the scope, carrying each interim and final report forward from the last one, setting the relevant policy clauses beside the facts, assembling business interruption claims and drafting claimant updates. Airclerk implements these as governed workflows connected to the records the firm already keeps. The adjuster reviews every draft and signs the report.
It can draft a likely cause for the adjuster to check, with a stated confidence and the evidence that would confirm or rule it out. On cover it goes no further than setting out the relevant clauses of the wording on file and the facts that bear on each. The adjuster gives the opinion and the insurer decides cover. Nothing the workflow drafts reaches a claimant without the adjuster's approval.
It is the normal case, and the design phase plans for it. The workflow drafts into the template and reporting guidelines of the insurer that instructed the claim, so the same firm can produce one insurer's format this morning and another's this afternoon. Which templates are built first is settled in design, usually by instruction volume.
No. The claims system stays the record of the claim. The workflow reads from what the firm already keeps, such as the claims system, email, photo storage and document folders, and works on what is still done by hand: the report narrative, the quote comparison and the chasing. Whether it connects directly or works from an export depends on what the system offers and what the firm's licence allows, and it is settled in the design phase.
Two controls come first. Claim isolation: one claim's file, and one instructing insurer's material, is walled off from another's. Permission-aware access: the workflow sees only the records the person running it is already authorised to see in the firm's existing systems. Before any claim information is connected, Airclerk reviews the AI vendor's enterprise terms, retention, training use, subprocessors and data-location arrangements, and checks them against what the firm's agreements with its instructing insurers require. The implementation runs on the firm's approved enterprise instance, under its own agreement with the AI provider.
In plain terms: the workflow keeps a record of its sources, the checks it ran and who approved the result. Airclerk calls this AI Process Assurance. A Workflow Charter defines the expected process, approved sources, professional boundaries and review gates. A Semantic Audit records what actually happened during the run. The retained AI Workpaper shows the sources used, exceptions raised, review changes and final approval. It does not claim to capture everything an adjuster does. It creates a reviewable record for the specific AI-assisted workflows the firm chooses to govern, and it is optional: the workflow runs without it.
Independent loss adjusting firms, in-house assessing teams inside insurers, and risk survey practices that report to underwriters. The first workflow is chosen around the firm's actual reporting rather than its label. If you are an insurer looking at the claims file from the other side, the insurers and MGAs page covers claims file summarisation and the rest of the underwriting and claims work.
Start with the question we would ask first: after a site visit, what do you still have to type up or chase before the report goes to the insurer? An introductory conversation is enough to say whether that is a first workflow and what it would cost to build.