Legal
How Janus Labs handles your information — including data processed by AI — in alignment with the Privacy Act 1988 (Cth) and the Australian Privacy Principles.
This Privacy Policy (the “Policy”) explains how Janus Labs Pty Ltd (“Janus Labs”, “we”, “us”, or “our”) handles your information when you interact with our company, website, or products. As an Australian technology company, we are bound by the Privacy Act 1988 (Cth) and the Australian Privacy Principles (APPs). We are committed to protecting your privacy and ensuring that our use of information — including personal information and data processed by artificial intelligence (AI) — is transparent, secure, and respectful of your rights.
This Policy is designed to be a “layered” notice, structured into two distinct parts:
By using our products, websites, or services, you acknowledge that you have read and understood this Policy and how it applies to your interaction with us.
To ensure clarity, we use specific terms in this Policy as defined under the Privacy Act 1988 (Cth) and general technology standards:
Janus Labs collects information only where it is reasonably necessary for one or more of our functions or activities. This principle of “limited collection” is central to our compliance with Australian Privacy Principle 3.
We categorise the information we collect as follows:
In Australia, the Privacy Act 1988 (Cth) focuses on ensuring that collection is for a “primary purpose” reasonably necessary for our functions or activities (APP 3). We do not currently offer products to users in jurisdictions that require a separate “lawful basis” framework (such as the GDPR in the European Union). When a Janus Labs product launches in such a jurisdiction, this section will be updated to record the lawful bases relied on; in the interim, where a separate lawful basis is required by an overseas user’s local law, Janus Labs relies on:
Janus Labs follows Australian Privacy Principle 6 regarding the use and disclosure of personal information.
Janus Labs does not sell your personal information to third parties. We do not trade your data for profit, nor do we provide it to data brokers.
To deliver our products, we use a small number of trusted third-party service providers. We only disclose information to these “processors” to the extent necessary for them to perform their functions, and they are contractually bound to maintain the confidentiality and security of your data.
Current processors:
ap-southeast-2 (Sydney) region.Planned processors (not yet engaged):
We will also update this list before introducing any new category of processor (for example, an analytics or payments provider).
We may disclose information if required to do so by a court order, subpoena, or a lawful request by an Australian government agency, provided such a request is valid and complies with the law.
Janus Labs aims to keep data local wherever possible, in alignment with Australian Privacy Principle 8.
By default, your data is hosted in Australia. We utilise Supabase as our primary database provider, with
data stored in the ap-southeast-2 (Sydney) region. This ensures your information remains within
the Australian jurisdiction for primary storage and backup.
No cross-border processing currently occurs. All Janus Labs processors that handle user data are Australia-resident, as listed in §5.2.
We are designing one optional future feature — the Tankful “your week” LLM-generated weekly summary — that may use a large language model provider whose processing region depends on which provider is selected. That feature is not yet enabled in any release. Before it is enabled, this policy will be updated in a versioned change to:
Until that update lands, no user data is sent overseas as part of any Janus Labs product.
Janus Labs uses computer programs and AI models to deliver innovative product features. We are committed to transparency regarding how these technologies affect you, as outlined in APP 1.7 and 1.8.
We use computer programs to summarise complex data, calculate performance scores based on telemetry, and provide personalised recommendations or coaching tips.
APP 1.8 (effective 10 December 2026) requires our privacy policy to disclose where a decision made by, or substantially contributed to by, a computer program “could reasonably be expected to significantly affect the rights or interests of an individual.”
Our current assessment is that the automated outputs generated by our products — such as driving scores or weekly summaries — are informational only. They are provided to the user for their own insight and improvement, and do not result in legal consequences, denial of services, or financial impact initiated by us. The product appendices below set out, for each product, the kinds of personal information used by computer programs, the kinds of decisions made solely by a computer program, and the kinds of decisions substantially contributed to by a computer program.
We review our AI assessments whenever product functionality changes or if data is shared with a third party (such as an insurer) who might use the output to make a significant decision. If a decision is made solely by a computer program that significantly affects you, we will ensure there is a path for you to request a human review.
Protecting your information is a core operational requirement for Janus Labs. We align with Australian Privacy Principle 11 and use industry-standard technical and organisational measures.
Janus Labs has adopted the Australian Cyber Security Centre (ACSC) Essential Eight Maturity Level 1 as our baseline target for internal systems. This includes:
Janus Labs does not keep information longer than is necessary for the purposes for which it was collected or to comply with legal requirements.
Retention schedules vary by product. For example, raw telemetry is typically deleted after a short period, while high-level aggregates may be kept longer to show you your history over time. When information is no longer needed, we either delete it or de-identify it so that it can no longer be linked back to you.
You have several rights regarding your personal information under Australian Privacy Principles 12 and 13.
You have the right to request access to the personal information we hold about you and to ask us to correct it if you believe it is inaccurate, out-of-date, or incomplete.
While the Privacy Act 1988 (Cth) does not impose a general right to erasure, Janus Labs provides deletion capabilities as a core privacy feature in our products. You can request the deletion of your account and associated personal information at any time, and we will delete the data we hold unless we are required by law to retain it.
Tankful (formerly known as Slipstream) is one app under both names — the deletion process is identical regardless of which name the user knows it by.
Primary path — in the app. If the app is still installed, open Settings → “Delete my account”. This is the fastest path and typically completes in under 10 seconds. It deletes everything Tankful holds about you both on the device and on our servers: drives, refuel records, scores, app caches, consent state, and the server-side uploaded rows associated with the active upload receipts on that device.
Lost-device or uninstalled path — by email. Email alex@januslabs.com.au with the subject “Data deletion request — Tankful”. By design, Tankful never stores an email address, phone number, advertising ID, Play account ID, or any other persistent server-side identifier that could be linked back to you — uploaded rows reference per-upload UUIDs whose only mapping exists in the local upload-receipt store on your device. Once that device is gone, there is no identifier left that lets us (or anyone else) tie server-side rows back to you. The practical deletion path is therefore the retention rules in §9 and Appendix A.3:
We respond to email deletion requests within 30 days in line with APP 12, and will confirm in writing that there is no linkable record of you on our servers (or, if there is one, that it has been deleted).
What cannot be deleted. Anonymous aggregate refuel statistics (postcode + fuel grade + day, k ≥ 10) persist indefinitely because they are not linked to any individual and cannot be re-identified. Removing one row from those aggregates would compromise the k-anonymity guarantee for everyone else in the same cell. See Appendix A.3.
If you have a concern about our privacy practices, please contact us first. We will investigate and respond within 30 days. If you are not satisfied, you may escalate your complaint to the Office of the Australian Information Commissioner (OAIC) at oaic.gov.au/privacy/privacy-complaints.
Our products are not directed at individuals under the age of 18. We do not knowingly collect personal information from children. If we discover we have collected information from a minor, we will delete it immediately.
Janus Labs complies with the NDB scheme under Part IIIC of the Privacy Act. If an “eligible data breach” occurs that is likely to result in serious harm, we will notify the Australian Information Commissioner and the affected individuals as soon as practicable, and no later than 30 days after becoming aware of the breach.
Changes: We post updates to our website (januslabs.com.au) and provide in-app
notifications for material changes.
Contact:
Janus Labs Pty Ltd
ABN 44 695 483 017. ACN 695 483 017. Registered in NSW, active since February 2026.
PO Box 530, Collaroy Beach NSW 2097, Australia
Email: alex@januslabs.com.au
Platform: Android (phone-only) · Region: Australia (NSW)
Tankful is an Android-based driving coach designed to help users improve road safety and fuel efficiency. It is built as a “local-first” application, meaning high-frequency sensor processing happens on your device and no backend account is required.
Telemetry in Tankful is strictly OPT-IN. Users must enable specific consent streams in the application settings:
| Consent stream | Streams uploaded | Data collected | Purpose |
|---|---|---|---|
consent_drive |
drive_features(and tank_summaries, derived) |
Per-drive aggregated features: distance, duration, idle fraction, cold-start duration, a behaviour proxy, predicted fuel, drive score, day-truncated drive date, vehicle class bucket, vehicle make bucket (rare AU makes collapsed to “Other”), vehicle segment bucket, vehicle ID hash, model and app version. Tank summaries pair drives between two refuels with predicted vs actual fuel for the tank. No per-second GPS, no sample-level data, no exact times — only day-resolution dates. | To generate per-drive efficiency scores, weekly summaries, and tank-to-tank model validation. |
consent_refuel |
refuel_events |
Per-refuel events: postcode (4-digit), day-truncated refuel date, fuel grade, litres, paid amount in cents, brand (or null), vehicle class bucket, vehicle make bucket (rare AU makes collapsed to “Other”), vehicle segment, vehicle ID hash, model and app version. | To compute tank-to-tank fuel economy, support regional fuel price aggregates (with k≥10 cell suppression — see §A.3), and refine the fuel model. |
consent_crash |
crash_reports |
Sanitised error reports: error class enum, screen enum, device class enum, and the minute (not second)
the error occurred. No stack traces, no exception messages, no free-form text fields.
Unknown error classes are bucketed to Other on device. |
To identify and resolve application stability issues. |
consent_retention |
retention_cohort_counts |
Weekly cohort counts: week ID, app version, and count of days active in that week. No per-row token, no device identifier, no per-user link. | To measure long-term retention and feature-launch impact at cohort scale. |
To link your drives over time for your own history without identifying you, Tankful creates a Vehicle Identity Hash using a salted SHA-256 algorithm. The salt is generated uniquely on your device and is NEVER uploaded to our servers. This ensures Janus Labs cannot reverse the hash to identify your device or link your data to other services.
We use automated database jobs (pg_cron) to enforce the following retention rules:
refuel_events, drive_features, tank_summaries,
crash_reports).This section satisfies the disclosure requirements set out in the new APP 1.7 / 1.8 (effective 10 December 2026). It is complete in v1 of this policy so the disclosure does not change as the deadline approaches.
Tankful’s automated features use only telemetry you have opted into:
drive_features (gated by consent_drive): distance,
duration, mean speed, idle fraction, cold-start seconds, behaviour proxy (pke_proxy_p95),
predicted fuel, drive score, day-truncated drive date, vehicle class bucket, vehicle make bucket, vehicle
segment bucket, vehicle ID hash, model and app version.refuel_events (gated by consent_refuel): postcode, refuel
date, fuel grade, litres, paid amount, brand, vehicle class/make/segment buckets, vehicle ID hash.tank_summaries (gated by consent_refuel): predicted vs
actual fuel for the tank, distance, drive count, duration, vehicle buckets and ID hash.The following Tankful outputs are produced entirely by a computer program, with no human in the loop:
drive_features
using a fixed, on-device heuristic (DriveScoreCalculator) that weights smoothness (40%), idle
fraction (30%), and sweet-speed adherence (30%). The score is shown to you on the drive detail screen and
contributes to your weekly summary.VehicleIdHasher on your device.The following outputs combine an automated step with a human-authored content layer:
VehicleIdentityMapper proposes a make and segment bucket from the human-readable display name
you choose; you confirm or override.APP 1.8 (effective 10 December 2026) requires this policy to disclose where automated decisions could reasonably be expected to significantly affect the rights or interests of an individual. Our current assessment for each Tankful automated decision is:
| Decision | Significantly affects rights/interests? | Reasoning |
|---|---|---|
| Per-drive score | No | Informational only. Shown to you. Not shared with insurers, employers, or any third party. Not used to set price, gate access, or trigger any consequence. |
| Weekly summary text | No | Informational only. Shown to you. |
| Vehicle ID hash | No | Linkage device for your own analytics; not a decision in the APP 1.8 sense. |
| Coaching tips | No | Suggestions from a human-curated library; you ignore or act on them. |
This assessment will be reviewed and this policy updated if any of these outputs ever leave your device for a third party (for example, an insurer pricing engine, a fleet operator dashboard, or a partner integration). If that ever changes, you will be re-prompted before any new disclosure begins.
The “Your week” screen in the current Tankful release shows computed numeric summaries (total distance, mean efficiency score, idle minutes, fuel cost) — no large language model is invoked, and no Tankful data is sent to any LLM provider in any current release.
We are designing an optional future enhancement in which a large language model would turn week-level aggregate statistics into prose. That enhancement is not yet enabled in any release. Before it ships, this policy will be updated in a versioned change to specify:
Until that update lands, the Tankful user experience contains no AI-generated text and no LLM call.
Additional product-specific appendices will be added here as new services are launched.