Real Estate Agent Tenant Reference Check Records: Use case and AI Confidentiality

After the applications close, the property manager phones the previous agent, the employers and the guarantor, and then has to turn a morning of calls into a recommendation for the owner. Pasting the call notes into ChatGPT or Claude to tidy them and rank the applicants is a natural shortcut. The notes hold more than a rental history. They hold what a referee said without being asked, an employer’s remark about leave, bank statement detail, a guarantor’s private doubts and a rival applicant’s pay. This page marks every item in a routine reference check record and shows how the same record is handled on Ermos inside the agency.

The tenant reference check

Completed by the property manager once a rental listing has closed and before the owner chooses. NSW Fair Trading’s guidance says an agent should collect only the minimum information reasonably necessary, which includes confirming identity, showing the tenants can pay the rent and showing they are likely to look after the property. It lists a reference from a previous landlord or agent as an example, and says unsolicited information that is not reasonably necessary should be destroyed as soon as possible.

Tenant reference check record
· Bronwyn Pike, Property Manager, Wattlebird Property

Applicants
Previous agent
Volunteered by the previous agent
Employers
Bank statements
Tenancy database
Guarantor
Other applicant
Assessment

An invented reference check record. The agency, the people and the employers are fictional and the telephone numbers are placeholders. Hover or tap any highlighted item for the analysis; the full list is below.

Reference check record preparer

An Agent+ that turns the property manager's call notes and replies into a tenant reference check record. It follows your own record template, keeps answers to the questions asked in one section, and flags unrequested remarks, such as personal or police detail, for the property manager to review against your own policy.

What it does

  1. Reads the call notes and written replies from referees
  2. Fills the record in your own template
  3. Flags unrequested remarks and gaps for the property manager

What stays with your people. The property manager decides what stays on the record and what is destroyed, and the owner chooses the tenant. The Agent+ prepares; it does not choose.

An Agent+ is built once and anyone in the practice can run it. It works from your own documents and runs on Ermos Dominion, the AI unit in your office, or on Ermos Edge, your own instance in an Australian data centre.

What leaves the agency

  1. 1
    The property, the rent and the closing date

    The address, the weekly rent and the date applications closed. The address and rent were on the advertisement for the rental and none of it is about a person, so little here depends on where it is sent.

  2. 2
    The applicants, their birth dates and the authority

    Two named people with dates of birth. Personal information. The authority they signed is what allows the checks, and it does not widen what the agency may keep or send on. NSW Fair Trading says only the minimum information reasonably necessary should be collected, and where the Privacy Act applies APP 3.2 limits collection to what is reasonably necessary for the agency’s functions or activities. Privacy Act guide.

  3. 3
    The previous agent’s reference

    A third party’s account of a named tenant’s payments, inspections and bond. Fair Trading lists a reference from a previous landlord or agent as an example of what is reasonably necessary. Where the Act applies, APP 3.6 says an organisation collects personal information about an individual only from that individual unless it is unreasonable or impracticable, so the agency should be able to say why the call was needed. Privacy Act guide.

  4. 4
    A detail the referee volunteered

    A police attendance and a domestic dispute involving the applicant and a former partner who is not an applicant, offered without being asked. Fair Trading says unsolicited information that is not reasonably necessary should be destroyed as soon as possible, and where the Act applies APP 4.1 and 4.3 ask for a decision within a reasonable period and destruction or de-identification if it could not have been collected. Pasting it into a model makes a copy before that step happens. Privacy Act guide.

  5. 5
    The employer checks

    Salary, tenure, a remark about leave and the co-applicant’s changed hours with a sum worked out from them. Income is personal information and Fair Trading treats it as part of showing the tenants can pay. A reason for leave can be health information, which is sensitive information, and APP 3.3 needs consent for it where the Act applies. The $36,600 is the manager’s own arithmetic on a contract that is not signed. Patient data guide.

  6. 6
    The bank statements

    Fair Trading says an applicant who has no payslips and gives bank statements instead should be encouraged to redact the daily transaction history and show evidence of savings. A statement with every transaction shows far more than the ability to pay, here a gambling pattern, and every line travels with the paste. APP 11.2 asks for reasonable steps to destroy personal information the agency no longer needs, where the Act applies. Privacy Act guide.

  7. 7
    The tenancy database search

    A result tied to two names. Fair Trading says agents may search a tenancy database and that it is best practice to tell applicants. A tenant can be listed on a database only after the tenancy has ended, only if named on the agreement, and only for a debt over the bond or a Tribunal termination order (Part 11 of the Residential Tenancies Act 2010). A clear result is thin, but it is still information about named people.

  8. 8
    The guarantor and her doubts

    A third party who is not an applicant, her income, her pension plans and a confidence about one applicant. Personal information about someone who did not apply, given to the agency for one purpose. Privacy Act guide.

  9. 9
    Another applicant’s details

    The rival’s name, occupation, pay and references sitting beside the first applicants. Personal information about a person who is not part of this check, and one who would not expect it to be compared in an outside service. Privacy Act guide.

  10. 10
    The assessment and the owner’s wish

    A ranking of people, the owner’s instruction and percentages worked out from the figures above. Fair Trading says it is unlawful in NSW to discriminate in providing accommodation on grounds that include race, sex, marital or partner status, age, disability, sexual orientation and gender identity, and that consideration should be given to whether any program or algorithm used by a third-party platform to assess an application could be assessing information in a discriminatory way. The recommendation is the property manager’s to make and explain, and the owner decides.

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 agency 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 property 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 EdgeErmos Dominion
Where the document is processedProvider servers, generally offshore on personal plansYour instance in an Australian data centre; answers generated in AustraliaThe unit in your office
Who can read itProvider staff and systems, under policyYour staff, within the boundaries you set. The named Australian provider processes only permitted excerpts, under contract, and keeps nothingYour staff only
Is it keptOn the provider’s termsKept on your instance; the provider is barred from retaining itStays on the unit
Can it train a modelOn consumer tiers, yesNo, by contractOnly your own unit learns from it, on your hardware

Part of Use cases in real estate agencies.

Common questions

Do ChatGPT and Claude use the information in a tenant reference check record?
Whatever is pasted or uploaded from the record is sent to the AI company's servers and processed there to produce a reply. 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. This record holds a police attendance the previous agent volunteered, a gambling pattern from bank statements and a guarantor's private doubts, alongside a rival applicant's pay and references.
Can we use ChatGPT or Claude for a tenant reference check record if we remove the names?
Not safely, because the record describes two applicants through their employers, their referee, their guarantor and their hours. 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. Income and a previous agent's account of a named tenant are personal information, and a reason for leave can be health information, which is sensitive information needing consent under APP 3.3 where the Act applies. 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.
Does deleting the chat remove a tenant reference check record from ChatGPT or Claude?
Not on its own. Deleting a chat removes it from the user's history, but OpenAI and Anthropic each say they can keep a deleted chat on their systems for up to 30 days, and longer where safety or legal obligations apply (OpenAI, Anthropic). NSW Fair Trading says unsolicited information that is not reasonably necessary, such as the police detail a referee offered unasked, should be destroyed as soon as possible. Deleting that line from the file afterwards does not undo a paste that has already made a copy elsewhere.
How can a property manager use AI for a tenant reference check record 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 summary of the checks is drafted from the application file and the agency's own template. The property manager checks every figure against the source, makes the recommendation, and leaves the choice to the owner.
What can a NSW agent collect to assess a rental application?
NSW Fair Trading's Commissioner's guidance says agents should collect only the minimum information reasonably necessary: enough to confirm identity, to show the listed tenants can pay the rent and to show the tenant is likely to look after the property. A reference from a previous landlord or agent is an example. Where an applicant has no payslips and gives bank statements instead, it suggests asking them to redact daily transactions and show savings, and for ID it suggests sighting documents rather than keeping copies.
What should an agent do with information a referee volunteers?
Fair Trading's guidance says unsolicited information, meaning information the agent took no active steps to collect, should be assessed, and if it is not reasonably necessary it should be destroyed as soon as possible. Where the Privacy Act applies, APP 4.1 and 4.3 say much the same: decide within a reasonable period whether the information could have been collected, and if not, destroy or de-identify it where lawful and reasonable.
Is there a standard rental application form in NSW?
The NSW Government said on 25 September 2026 that Parliament had passed the Residential Tenancies Amendment (Protection of Personal Information) Bill 2025, and the NSW Parliament bill page records assent on 2 October 2026 (Act 38 of 2026). The Government said the new rules will require a standard application form and limit driver licences and passports to the preferred applicant before the lease is signed, and are expected to come into effect in early 2027. It is consulting on the form, so check the NSW Government site for the start date.

See Dominion answering from your own documents.

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

Book a demo