You do. Raw sensor readings (soil moisture, temperature, salinity, pump on/off events) are facts about your land. Under U.S. copyright law (Feist Publications v. Rural Telephone Service, 499 U.S. 340 (1991)), raw facts can't be copyrighted, and no one, including us, can claim to “own” them. We hold them for you, process them on your behalf, and license them only as you direct.

What we do own is the derived analytics: the AI model weights, the prediction outputs, the benchmarks, and the dashboards we build on top of your data. Same as a CPA owns the spreadsheet, but you own the underlying receipts.

No. Not your lender, not your crop insurer, not your buyer, not your packing house. We do not ship your raw sensor data, your pump records, your GPS boundaries, or any individually identifiable account information to any third party for any commercial purpose without your written authorization for that specific recipient and purpose.

If we ever build a financial data product (for example, a basin-level risk index sold to ag lenders), it will use only aggregated, anonymized data with at least seven (7) unaffiliated farms in any reported cell, county-level geography or broader, and no GPS, no APNs, no field boundaries. You can opt out of having your data contribute to those aggregated products. See Section 8 below.

Honest answer: we are still getting written confirmation, and we will update this page when we have it.

Here's what we know today. Sentek's publicly posted Terms & Conditions of Trade (§13) cover hardware IP and Sentek's own development work; the most natural reading does not give Sentek ownership of your sensor readings or of derivative products built from them. Sentek does not publish a public IrriMAX Live End-User License Agreement, Subscriber Agreement, API terms, or data-ownership policy, which means the precise data-handling terms of the IrriMAX Live cloud platform are not on the public record. We have asked Sentek in writing to confirm: (a) the customer owns the raw sensor data flowing through IrriMAX Live, and (b) Sentek does not access or use customer data for its own commercial AI/ML purposes.

Until we have that letter, we treat IrriMAX Live as a potential third-party data path and we are actively standing up direct SDI-12 / Modbus RS-485 ingestion that pulls raw data straight from the probe to our cloud, bypassing IrriMAX Live entirely.

The short version: in BYOS, you control the sensor manufacturer's relationship, and we are only the analytics layer. Whichever path you choose, AgWaterAI's commitments on this page apply equally. BYOS does not give us a smaller obligation; Managed does not give us a bigger right.

Yes. You can request a full export of your raw sensor data, your pump records, your dashboard inputs, and the predictions we have generated for you, in standard formats (CSV, Parquet, or JSON). We will fulfill the request within thirty (30) days of a written request to data-officer@agwaterai.com.

What stays with us after you leave: the trained AI model weights themselves. Neural-network and gradient-boosted model weights are a one-way mathematical function: once your data is incorporated into a trained model, the individual readings cannot be reconstructed from the weights. Under Feist, the raw facts you contributed are not copyrightable in the first place, so we are not asserting any retained license over your raw data; we simply cannot un-train a model.

Yes, with three honest exceptions. On written request, we will delete: your account, login credentials, and dashboard; your raw sensor readings from our long-term storage; your GPS coordinates, APNs, and field boundary data; your personally identifiable contact information.

We may retain, for specific defined purposes:

1. De-identified aggregated metrics that have already been incorporated into trained model weights, where individual re-identification is no longer mathematically possible.
2. Submission metadata (IP, user-agent, timestamps) required to support active fraud investigations.
3. Records of payments and contracts we are required to keep under California tax law and IRS recordkeeping rules (Cal. Civ. Code § 1798.105(d)(7) permits this carve-out).

We will not retain identifiable raw sensor readings after a verified deletion request.

Yes, for paid Pro / Managed subscribers, we use your sensor data to train and improve the production ML models behind our six decision systems (PULSE, SENTRY, SENTRY-TRANSITION, SALT FLUSH, WATCHDOG, and the irrigation propensity model). This is how we keep per-acre pricing where it is. We disclose this in our Terms of Service and in our general Privacy Policy.

You can opt out of model training without losing access to the platform, consistent with the CCPA's non-discrimination rule (Cal. Code Regs. tit. 11, § 7080). If you opt out, your dashboard, your predictions, and your SGMA reports keep working, but your data is processed in an ephemeral pipeline that does not contribute to model retraining or to aggregated products.

To opt out, email data-officer@agwaterai.com with subject line “Opt out of model training.” We will confirm within ten (10) business days.

For our free tools (SGMA Allocation Estimator, penalty estimator), we do not train models on what you enter. Inputs are used to generate your result and then dropped from long-term storage.

No, not without your written authorization for the specific paper or study.

We have a long-term interest in publishing validation work with the University of California Cooperative Extension and similar academic partners. When we do that work, we ask each participating grower for written consent for the specific study, identifying the researcher, the dataset scope, the publication venue, and the destruction date for the working copy. Anything else falls under your standard “no third-party sharing” protection.

We additionally commit that any data we share with academic partners will be k-anonymized to k ≥ 5 at the county-and-crop level (no county-crop combination with fewer than five unaffiliated farms), with GPS truncated and acreages bucketed, consistent with the Cal. Civ. Code § 1798.140(m) “de-identified” standard.

We will not voluntarily disclose your raw farm data to any government agency, including your Groundwater Sustainability Agency, the State Water Resources Control Board, the California Department of Food and Agriculture, the U.S. Department of Agriculture, or county-level enforcement bodies, for regulatory enforcement purposes.

What we will do, because we have to:

• Comply with a valid subpoena, search warrant, or court order. A clause in a privacy policy promising “we will never share with government” is void against a valid court order.
• Notify you promptly of any government request for your data, unless we are legally prohibited from doing so, so you have the opportunity to seek a protective order from the court.
• Require valid legal process. We do not respond to informal agency inquiries, “voluntary cooperation” requests, or fishing expeditions.

A practical note on California-specific water law: California Water Code § 10730.8(b) allows GSAs to require well-extraction data as part of compliance, but the duty to provide that data runs from you to your GSA, not from us to your GSA. Your MSA with us authorizes us to help you prepare and submit that data on your behalf at your direction. We do not push it to your GSA without your sign-off.

We notify you in writing within 72 hours of confirming a security incident that has materially compromised, or is reasonably believed to have compromised, your sensor data, account credentials, or personally identifiable information. The notice will tell you what we know, what we don't yet know, what data categories were affected, and what we are doing about it.

You have a private right of action under California Civil Code § 1798.150 if a breach of certain categories of personal information results from our failure to implement reasonable security. We carry cyber-liability insurance against this risk. We maintain industry-standard controls: TLS in transit, AES-256 at rest, principle-of-least-privilege access, multi-factor authentication on all admin accounts, and an annual third-party security review.