Trials Within Reach mark An open teal loop with a rising path and a teal point representing opportunity within reach. Trials Within ReachPublic trial discovery

Effective date: 2026-11-17

Privacy — Trials Within Reach Public Discovery

Publisher: Trials Within Reach

Privacy and support contact: udit.nindra@trialswithinreach.com

Scope

This notice covers the TWR public-discovery plugin and the information pages on plugins.trialswithinreach.com. The plugin retrieves published public trial information. It does not link to a TWR account or provide access to MySite, private professional information, patient records, finance or site operations. All callers have the same public scope, including clinicians and TWR staff.

What reaches the service

ChatGPT sends the selected public search choices: country, specialty, published condition, phase, recruitment status, registry identifier and published study site or service area. A detail request uses a public study identifier. A request for another page includes a temporary pagination reference.

The tools have no field for a patient profile, clinical narrative, name, date of birth, medical record number, home address, GPS location or clinical upload. Do not send those details, passwords or tokens through this service or in a support report. Strict inputs reduce risk; they do not guarantee that a conversation has been de-identified.

We use the limited search choices to retrieve and filter approved public catalogue records. Our public sources are hosted by Supabase and the TWR website. Upstream requests use fixed public operations and the public country, specialty or study/site identifiers needed by those operations. We do not forward your TWR login, cookies, staff tokens or full conversation.

The plugin backend makes no model-provider or embedding calls. ChatGPT separately processes your conversation under its own policies and your account settings. This notice does not replace the privacy notices for ChatGPT, the separate TWR website or external registries you visit.

What is stored, and for how long

The TWR application processes tool requests in memory. It does not maintain a search-history, patient-profile or conversation database. Pagination references expire after 15 minutes. Rate controls use temporary in-memory hashes of the connection address; request counts, timings, response sizes and error counts are also held in process memory. Application logs omit request bodies, search values, medical text, credentials and raw upstream error messages.

Amazon Web Services (AWS) hosts the service in Sydney, Australia (region ap-southeast-2), on a single container behind an AWS load balancer. The load balancer's access logs and connection logs are switched off. The container log receives only a start-up line and is kept for 14 days. AWS still processes connection metadata, such as the source IP address, to deliver and protect the service; for plugin requests the source IP may belong to ChatGPT or a proxy, and when you visit these pages directly it may be your own IP address. No external log stream is configured. The 14-day log period is not a promise of erasure from every provider backup or other provider record.

Tool arguments use request bodies, not URL query strings. We have not established a provider-level guarantee that bodies are never captured. Infrastructure may receive or record information before an invalid request is rejected. Rejection does not erase a provider record. Do not put sensitive information in URLs or headers.

These information pages contain no TWR analytics scripts, advertising trackers, forms or application cookies. Hosting providers still process connection metadata. You can choose not to use the plugin or these pages; you can instead visit the public registries directly.

People and providers involved

Authorised TWR operators use necessary records to operate the service, diagnose faults, protect it from abuse and answer support or privacy requests. Udit Nindra is the support and privacy lead. Relevant providers are AWS, our email-hosting provider, and the providers hosting TWR's public catalogue sources. Provider staff may process information under their service and support arrangements. The Sydney service location does not establish the location of every provider's support or processing system.

We do not use search inputs or operational records for advertising, personal profiling or clinical eligibility decisions. We limit any diagnostic exports to what is necessary for the issue. Routine support correspondence and exported diagnostics are removed from TWR's active mailbox or working records within 90 days after the case is closed. Necessary legal or security records may be kept longer, with the reason and retention period documented. Provider backups, retention controls and records held by other services may follow different periods; we do not promise complete or immediate deletion from those systems.

Support, accidental input and privacy requests

If you email us, our email provider receives your address and the information you choose to send. We use it to investigate and reply. Do not include clinical records, complete chats, screenshots containing personal information, passwords or tokens. For accidental disclosure, give the approximate time and the channel used without repeating the sensitive contents.

We assess the report, minimise further copying or sharing, restrict access where needed and seek provider assistance if necessary. We explain the steps available and any limits on containment or deletion. We do not reuse accidentally supplied clinical information for trial matching.

For access, correction, deletion or a privacy complaint, contact the address above. We may ask for the minimum information needed to verify and locate the request; do not send identity documents unless a necessary, secure process has been agreed. We review the request, explain our response and any applicable limits, and provide the relevant complaint-escalation information. We do not promise a particular response time or provider-log early deletion.

Disconnecting the plugin in ChatGPT stops new requests through that connection. It does not itself delete existing ChatGPT conversations, provider records or support correspondence; those have separate controls and retention rules.

Updates

We publish the current version and effective date on this page. Material changes will be highlighted here and in the plugin release notes. Check this notice before continuing to use the service after an update.