Needs analyses and AI data leakage: what a client fact find gives away in ChatGPT or Claude.

The needs analysis is where a broker learns the business: what it does, what it owns, who runs it, what it is contracted to deliver, and where it is exposed. It is the document that gets pasted into ChatGPT or Claude to be turned into a summary for the file or a submission for the market. This page marks every item in a standard commercial needs analysis that the Code of Practice and the Privacy Act bear on, and shows how the same analysis is written up on Ermos inside the brokerage.

The needs analysis

Completed with a new commercial client before the first placement, and updated at each renewal, to support advice in the client’s best interests.

Commercial needs analysis
Marchetti Dental Pty Ltd trading as Northside Smile Studio¹ · P. Desai⁹ · 15 October 2026

Directors	Lena Marchetti (DOB 3 July 1976, 8 Carrington Avenue, Mosman) and Marco Marchetti (DOB 19 November 1979). Both dentists. Marco has a 2022 AHPRA condition on his registration (completed supervision requirement, now lifted).²
Business	4-chair dental practice, turnover $2.1m, payroll $780,000 across 8 staff, patient records held in Dental4Windows on a server at the practice and backed up to a US cloud service.³
Property and equipment	Premises leased from the Marchetti Family Trust at 22 Industrial Drive, Brookvale (related party; the trust is considering selling in 2027). Equipment $610,000 including CBCT scanner ($185,000). Fit-out $240,000.⁴
Contracts and exposures	Preferred provider agreements with two health funds (volume $600,000). Sedation services provided by a visiting anaesthetist under contract. Three complaints to the Dental Council in five years, one upheld (2022, informed consent, which led to the supervision condition), no claims.⁵
Existing cover	Professional indemnity $20m (Dental Protection, held individually). Business pack with Meridian, renewal 1 December. No cyber cover despite the 2025 phishing loss of $14,000. No management liability.⁶
Gaps identified	Cyber (patient records, US cloud backup); management liability (employment practices, after a 2025 unfair dismissal claim settled for $18,000); business interruption sum insured not reviewed since 2022.⁷
Client priorities	Lena wants comprehensive cyber cover; Marco thinks it is unnecessary. Lena mentioned the partners are “not on the same page about a lot of things lately”.⁸

An invented needs analysis. Hover or tap any highlighted item for the analysis; the full list is below.

What leaves the brokerage

The client

Named. Everything below is identifiable. Privacy Act guide.

Frontier Models (ChatGPT, Claude)	Transmitted to the provider, generally offshore on personal plans
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

The directors, with dates of birth, an address and a registration history

Two identities, a home address, and a professional regulator’s condition on one of them. Personal information, and a professional history that is nobody’s business outside the placement. Privacy Act guide.

Frontier Models (ChatGPT, Claude)	Two identities and a regulator history transmitted
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

The business, its payroll and where patient records live

Turnover, payroll, and the fact that a dental practice backs up patient records to an overseas cloud service. Commercial information, and a security fact about health records.

Frontier Models (ChatGPT, Claude)	Payroll and a health-records security fact transmitted
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

Premises, equipment and a possible sale

A related-party lease, equipment values and the landlord’s intentions. Commercial and family information. Privacy Act guide.

Frontier Models (ChatGPT, Claude)	A related-party lease and a possible sale transmitted
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

Contracts, a visiting anaesthetist and regulator complaints

Health fund agreements, a named role under contract, and a complaints history with one upheld finding. The exposures the broker exists to insure, and a professional history that would damage the practice if it surfaced.

Frontier Models (ChatGPT, Claude)	A complaints history transmitted; on consumer tiers it can train future models
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

Existing cover and the uninsured phishing loss

What is insured and what is not, including a prior cyber loss with no cyber cover. The gaps, mapped.

Frontier Models (ChatGPT, Claude)	The gaps in cover transmitted
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

The gaps and an employment claim

Uninsured exposures, including a settled unfair dismissal claim with the amount. Information about a former employee’s dispute. Privacy Act guide.

Frontier Models (ChatGPT, Claude)	A settled employment claim transmitted
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

A disagreement between the partners

That the two directors disagree about cover and, in the client’s words, about a lot of things. A confidence that bears on the business’s future. Privacy Act guide.

Frontier Models (ChatGPT, Claude)	A disagreement between the partners transmitted
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

The broker

Ties the analysis to a named account manager.

Frontier Models (ChatGPT, Claude)	Transmitted, and the leak is traceable to you
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

What happens in Frontier Models (ChatGPT, Claude)

The whole document is transmitted to the provider’s servers. On personal plans the user cannot choose where it is stored or processed, and for the major providers that has generally meant outside Australia, under that country’s legal process. Some business plans offer storage in Australia, but the provider still receives the content.
On consumer tiers, prompts can be retained and used to train future models.
The provider’s staff and systems can reach the content under the provider’s own policies.
Once sent, the brokerage no longer controls it. Retention and deletion happen on the provider’s terms.

The same document on Ermos

Ermos Dominion

The document is drafted on a unit in your office, from your own templates and the client file. Every marked item stays on the unit, which keeps working with the internet switched off. Nothing in it is transmitted to an AI provider, so there is no disclosure to analyse: the confidentiality work is the same as for your file server.

Ermos Edge

Your documents, the index and your access boundaries stay on your own instance in an Australian data centre. When a staff member asks for this draft, only their question and the excerpts they are permitted to see go to a named Australian provider, who is contractually barred from keeping the content or training any model on it. Nothing is processed outside Australia.

	Frontier Models (ChatGPT, Claude)	Ermos Edge	Ermos Dominion
Where the document is processed	Provider servers, generally offshore on personal plans	Your instance in an Australian data centre; answers generated in Australia	The unit in your office
Who can read it	Provider staff and systems, under policy	Your staff, within the boundaries you set. The named Australian provider processes only permitted excerpts, under contract, and keeps nothing	Your staff only
Is it kept	On the provider’s terms	Kept on your instance; the provider is barred from retaining it	Stays on the unit
Can it train a model	On consumer tiers, yes	No, by contract	Only your own unit learns from it, on your hardware

Part of AI data leakage in insurance brokerages. See how this works in practice: Renewal preparation on Ermos.

Common questions

Do ChatGPT and Claude use the information in a needs analysis?

Both receive the whole analysis on the AI company's servers once it is pasted. As at October 2026, whether a conversation on a personal plan (ChatGPT Free, Plus or Pro; Claude Free, Pro or Max) can be used to train the models depends on a setting each user controls. Business plans (ChatGPT Business and Enterprise, Claude Team and Enterprise) and the developer APIs sit under commercial terms, and OpenAI and Anthropic each say they do not train on that content by default (OpenAI, Anthropic). On either product, the thumbs up and thumbs down buttons can send a whole conversation back for training. A needs analysis holds a director's home address, the payroll and the gaps in the practice's cover.

Is it safe to paste a needs analysis into ChatGPT or Claude if we remove the directors' names?

Not safely, because the rest of a needs analysis still points to two people. Removing names rarely makes a document anonymous. Information stays personal information while the person is reasonably identifiable, and the OAIC warns that information can be at risk of re-identification even when it has been de-identified or anonymised. Two dentists who run a four-chair practice, one with a past registration condition and both with a stated date of birth, are easy to identify. The Office of the Australian Information Commissioner (OAIC) recommends in its guidance on commercially available AI products that organisations do not enter personal information, and particularly sensitive information, into publicly available generative AI tools, because of the significant and complex privacy risks.

Is it safe to use an AI note-taker or meeting recorder for a needs analysis?

Not until the brokerage has checked where the recording goes, who can review it and whether it is used to train a model. The OAIC's guidance also tells organisations to review carefully the terms and settings that will apply to their use of any AI product. Clients speak freely in a needs analysis meeting, and a recorder captures every remark, including a disagreement between the partners such as the one about cyber cover. A note-taking tool is a separate product from ChatGPT or Claude, with its own terms.

How can a brokerage use AI for a needs analysis without sending it to ChatGPT or Claude?

Ermos is built so the document stays under your control. On the Dominion AI unit it is drafted inside your office and none of it is sent to an AI provider. On Ermos Edge your documents and index stay on your own instance in an Australian data centre, and only the question and the excerpts the user is permitted to see go to a named Australian provider that is contractually barred from keeping them or training any model on them. The analysis is drafted from the meeting notes, the existing policy schedules and the client's documents, in the brokerage's own format. The broker checks it, completes it and signs it off before it goes on the file.

Why does a needs analysis hold so much about the directors personally?

Because commercial insurance underwrites the people as well as the business: management liability, professional indemnity and the cyber questionnaire all ask about the directors' histories. The broker collects it to place the risk, and holds it under the Privacy Act.

The practice's own cloud backup is overseas. Isn't that the bigger problem?

It may well be, and it is the kind of thing a good broker flags, as this one did. It does not make the broker's own handling of the client's information less important. The brokerage holds the gap analysis; the brokerage decides where it goes.

How is a needs analysis written up on Ermos?

The meeting notes, the existing policy schedules and the client's documents are indexed on your own unit, and the analysis is drafted from them in the brokerage's format. The broker completes and files it. Nothing about the practice, its directors or its patients leaves the office.

See Dominion answering from your own documents.

A demonstration takes 30 minutes, uses no client data, and comes with no obligation.

Book a demo

Keep reading

AI data leakage in insurance brokerages AI for insurance brokers The Privacy Act, the APPs and AI All professions