Skip to content
LEGAL

Privacy Policy

Draft — pending legal reviewLast updated 3 October 2026

How FleetWarrant handles personal information: both the ordinary website/waitlist kind, and the Event data your Agents generate once you’re a Customer.

01

Scope, and two different roles we play

This Privacy Policy explains how FleetWarrant Pty Ltd [entity name and ACN to be confirmed on legal review] (“FleetWarrant”, “we”) handles personal information. It covers two different relationships, because we play two different roles depending on the data:

  • As a controller: for people who visit https://fleetwarrant.com, join our waitlist, or contact us. This is ordinary website-visitor and prospect data, and we decide how it’s used, as described below.
  • As a processor: for the Event data our Customers’ Agents generate once they’re on the Service. We handle that data on the Customer’s instructions, under our Terms of Service and their Order Form, not on our own. If you’re an individual whose information shows up inside a Customer’s Event data (for example, because you’re a colleague or counterparty an Agent interacted with), that Customer is who to contact about it; see Event data below for how it’s handled.
02

Information we collect from visitors and applicants

When you join the waitlist or otherwise ask to hear from us, we collect what the form asks for: your name, work email, company, role, a rough estimate of how many Agents you run, and anything you choose to add in the free-text field. We also capture, for that submission only, the page you came from and any campaign parameters in the URL, and your consent to be contacted.

The form is delivered to us by our forms processor, Anchor Forms. Submissions are screened for spam by a hidden field only a bot would fill in, an allowlist restricting submissions to our own domain, and rate limiting keyed to the submitter’s IP address. None of these read anything about you beyond what you typed into the form. We don’t run advertising or analytics trackers on this site (see Cookies).

If you email us or otherwise contact us directly, we collect whatever you send us and use it to respond.

03

Information we collect from customers

When you or your company becomes a Customer, we collect account and billing information: names, work emails, and (once self-serve billing is live) payment details handled by our payment processor, never stored by us in full. We also collect the information you give us about each Agent you register: its name, its purpose, the models it may call, and who created it.

04

Event data: what your Agents generate

Your Agents send their LLM-provider calls through the FleetWarrant Gateway instead of directly to the provider. The Gateway forwards each call on unchanged, and reports a record of it (an “Event”) to the Service. Three different things can happen to a call’s content, and this section keeps them apart:

  • In transit, we see every prompt and response in the clear. The FleetWarrant Gateway is the path your Agents’ traffic actually takes to reach an LLM provider, so it necessarily receives each request and response unredacted, to forward it. There is no client-side step that strips anything before it reaches us. Redaction applies to what we keep, not to what passes through; what we keep is described next.
  • By default, we keep a redacted preview, not the raw request and response. Unless a Tenant admin turns on raw capture (see below), FleetWarrant does not keep the raw request and response. For each call we store a redacted preview, encrypted under your Tenant’s own KMS key, plus the call’s metadata: which Agent and Connector, the model, token counts and cost, timing and status, and the Action it was classified as. To make the preview, the Gateway scans free-text message content for a fixed set of known, high-confidence patterns and redacts each wherever it appears:
    • AWS access key IDs: AKIA... (followed by 16 more characters)
    • FleetWarrant Agent API keys and registration secrets: fwk_... / fwreg_...
    • OpenAI API keys: sk-...
    • Anthropic API keys: sk-ant-...
    • Generic bearer tokens: Bearer <token>
    • US Social Security Numbers: 123-45-6789 (hyphenated only, not bare or space-separated digits)
    • Email addresses: jane@example.com
    • Credit-card numbers: 4111 1111 1111 1111 (13-19 digits, Luhn-valid runs only)
    It separately masks a fixed set of sensitive field names (such as password or ssn) anywhere they occur as a structured field. A copy of the preview is written to the audit archive too.
  • If a Tenant admin turns on raw capture, the full request and response are archived as well, and what the preview removed is still in them. Raw capture is off until an admin of your Tenant switches it on. While it is on, the complete request and response are written to the audit archive in addition to the redacted preview, each encrypted under its own per-Event key. A secret, Social Security number, email address or card number that the preview removed therefore remains in the encrypted raw copy; encryption protects it, but it doesn’t redact it. Turning raw capture off stops new raw bodies from being archived. It doesn’t remove the ones already archived, which stay until they’re shredded (see Storage, security, and retention). Through the Service, a raw body is read only by an evidence export that an admin of your Tenant asks to include raw bodies; the export is marked as holding unredacted data.
  • The redaction is not general PII or business-data redaction, and it isn’t configurable per Customer. The ruleset is fixed and applied identically to every Customer. There’s no way today to add your own patterns or fields. A name, address, tax ID, or other free-text personal or business information that doesn’t match one of the specific patterns named above (including a credential in a format the ruleset doesn’t yet recognise) will appear unredacted in the preview that we keep and show in the dashboard and, if raw capture is on, in the raw copy. Don’t route data you need redacted or withheld from us through the Service until that’s changed. Talk to us at matt@fleetwarrant.com if this is a blocker for you.

We act as a processor over Event data: we don’t use it to build profiles of the individuals it mentions, sell it, or use it for advertising. We may use de-identified, aggregated Event statistics that can’t reasonably be traced back to you, your Agents, or the individuals your Agents interacted with, to improve the Action Mapper’s pattern library and the Service’s reliability.

05

How we use information

  • to provide, operate, and secure the Service;
  • to respond to your waitlist request or enquiry, and to follow up about early access, where you’ve consented to that;
  • to bill and support Customer accounts;
  • to detect, investigate, and prevent abuse, fraud, or security incidents; and
  • to meet legal, tax, and regulatory obligations we’re subject to.
06

Storage, security, and retention

The Service runs on Amazon Web Services, in a single AWS region: Sydney (ap-southeast-2). We don’t currently offer a choice of region or a data residency commitment to any particular country or bloc. If your organisation needs one before you can use the Service, tell us at matt@fleetwarrant.com. Every Customer’s data is logically isolated by a tenant identifier on every read and write, and the payload-bearing parts of the audit record are additionally encrypted under an encryption key dedicated to your account, not shared with any other Customer’s. Each raw body, where raw capture is on, is also sealed under a key of its own.

The audit archive uses S3 Object Lock in GOVERNANCE mode. Each archive object is locked for 1 year by default, and you can change that period for your Tenant within the limits of your plan. The lock is a minimum, not a deletion date: our application never deletes an archive object, so an object can outlive its lock, still encrypted. It is deletion-resistant, not tamper-proof: no FleetWarrant application role can bypass the lock, but an administrator of the AWS account that hosts the Service holds a permission that can. The archive holds the redacted preview and the call’s metadata for every Event, and the raw request and response only for a Tenant that has turned on raw capture. Its value as compliance evidence is that records in it aren’t edited or deleted individually.

When your account is offboarded, we remove your data by destroying your account’s dedicated encryption key, which makes the archive unreadable, rather than by editing archived records one at a time. Destroying a key is not the same as deleting the encrypted objects, which our application doesn’t do.

Erasing one record inside an active account works differently for each part of it. A raw body can be shredded on request, or by your Tenant’s raw-body retention; a legal hold pauses shredding until the hold is released. Shredding destroys the body’s key, not the archived object: the body becomes unreadable on every live path at once, and unrecoverable from our backups once their 35-day window has passed. The redacted preview and the call’s metadata can’t be erased one record at a time. They stay in the archive, readable under your account’s key, for as long as the archive is kept. The lock period is a minimum, and our application never deletes an archive object, so in practice that means until your account is offboarded and its key is destroyed. We haven’t designed a per-record erasure for them. If you need this resolved contractually, tell us at matt@fleetwarrant.com before sending regulated data through the Service.

Waitlist and prospect data is kept for as long as reasonably needed to run the waitlist and follow up on it, or until you ask us to delete it.

07

Who we share information with

We don’t sell personal information. We share it with:

  • Infrastructure and security providers: Amazon Web Services (hosting and storage) and Cloudflare (hosting and content delivery), who process data on our instructions to run the Service.
  • Anchor Forms: processes waitlist and enquiry submissions on our behalf.
  • Payment processors: once self-serve billing is live, a processor such as Stripe handles your payment details directly; we don’t store your full card number ourselves.
  • Legal and safety: where required to comply with law, or to protect the rights, property, or safety of FleetWarrant, our Customers, or others.

We never disclose one Customer’s Event data to another Customer, and we don’t use it to train or fine-tune a general-purpose model outside the Service.

08

Your rights

Depending on where you’re located, you may have rights to access, correct, port, or request deletion of your personal information, to object to or restrict some processing, and to withdraw consent (which won’t affect anything we did before you withdrew it). To exercise any of these for information we hold as a controller (website visitor and waitlist data), email matt@fleetwarrant.com.

If your request concerns Event data we hold as a processor for a Customer, we’ll direct you to that Customer, who controls it, unless you tell us who they are and ask us to help directly.

09

International transfers

The Service runs in a single AWS region (see Storage, security, and retention), so if you or your organisation are located elsewhere, your information is necessarily processed outside your own country. For some visitors and Customers, that includes a transfer out of the EU/EEA or UK. We haven’t yet put a specific transfer mechanism, such as Standard Contractual Clauses, in place for every combination of visitor location and processing location this could involve. If your organisation needs one documented before you can use the Service, tell us at matt@fleetwarrant.com and we’ll work through it as part of onboarding.

10

Cookies

This site doesn’t run analytics or advertising cookies, and the waitlist form doesn’t currently use Cloudflare Turnstile or any other bot-check that sets one either. The form is protected instead by a hidden honeypot field, an origin allowlist, and per-IP rate limiting, all evaluated on submission rather than through anything stored in your browser. Our hosting provider, Cloudflare, may still set strictly necessary cookies to route and secure traffic to the site.

11

Children’s privacy

The Service is a B2B product and our website isn’t directed at children. We don’t knowingly collect personal information from anyone under 16. If you believe a child has given us information, contact us and we’ll delete it.

12

Changes to this policy

We may update this Policy as the Service and our practices change. For a material change, we’ll update the “Last updated” date above and, where the change is significant, give Customers notice by email or in-product notice.