Marvel Data Usage and AI Processing Policy

Effective date: September 28, 2026 · Version 3.1

This policy explains what data Marvel machines collect, what Keirton Inc. does with it, and the commitments we make about it. It covers machine and production data from Marvel units processing your product, in your facility or during a demonstration at Keirton. It does not cover information you submit through our websites; that is governed by the Keirton Privacy Policy, published on each site. The two documents do not overlap.

If you have a signed agreement with Keirton, such as the Marvel Master Agreement, that agreement governs. Where this policy and your agreement differ, your agreement controls.

1. What Marvel collects

During normal operation, the machine collects:

Flower images. Multi-angle images of each flower captured by the machine's cameras.
Grade results. The AI's classification of each flower: quality class, detected attributes such as stems, leaf, structure, color, and size, and the output stream it was routed to.
Machine settings and telemetry. Operator-defined thresholds and sensitivity settings, throughput rates, run times, and machine health data.
Batch and run records. Aggregate counts, weights where measured, and distributions for each run.

The machine does not collect personal information about your staff. Its cameras image flower in the inspection chamber, not people.

2. Where processing happens

Grading decisions are made on the machine, in real time. Every flower is imaged, classified, and routed by the machine itself, with no cloud connection required for grading. The machine connects periodically for software and model updates and to sync data; grading continues to work between connections.

After operation, a sample of flower images, together with grade results and machine telemetry, is transferred from the machine to Keirton's cloud infrastructure for the purposes in Section 3. Full image streams are not transferred.

3. What the data is used for

Operating your machine. Real-time grading and sorting to your settings, and the batch grade reports your machine produces.
Supporting and maintaining your machine. Diagnostics, service, and performance monitoring by Keirton's team.
Improving the AI. Collected data trains and refines the grading models, improving classification accuracy and consistency across the fleet. Improvements benefit every Marvel machine, including yours.
Publishing results. Fleet statistics and anonymous results, as described in Section 6.

Data is used for these purposes only. We do not sell machine or production data.

4. Who owns the data

Your production data is yours. The images, grade results, and production records Marvel captures from your product belong to you. Keirton receives a permanent, royalty-free license to use them for the purposes in this policy. Keirton owns Marvel's software and AI models, and everything it builds from the data.

5. Sharing

Apart from publication under Section 6, we do not share your data with third parties, except:

with service providers that host and process it for us, such as AWS, under confidentiality;
where the law requires it;
with your written consent; or
with a successor to Keirton's business, which must honor this policy.

6. Publication of results

Anonymized fleet statistics. Keirton may publish aggregated, anonymized statistics derived from machine data across the fleet, such as industry-level yield and quality benchmarks. Published statistics never contain customer-identifiable data, and no individual operation's data is disclosed, identifiable, or reconstructable from them.

Anonymous results. Keirton may publish results from a single operation, such as a case study or demonstration result, without identifying you. Before publication, we remove your name, location, and any other detail that could identify your operation, and we show you the draft.

Named results with your consent. Where you give written consent to a specific publication, such as a named case study or a customer-approved evaluation report, that publication may identify you and include your results. Consent is specific to the publication, you approve the wording before release, and nothing is published under your name without it.

If you want your data excluded from fleet statistics or anonymous results, contact us at the address in Section 10 and we will exclude it from anything published after we receive your request.

7. Security

All data transferred from the machine is encrypted in transmission.
Stored data files are encrypted at rest with dual-layer AES-256 encryption.
Data is stored on Amazon Web Services (AWS) infrastructure located in the United States.
Access is restricted to authorized systems and personnel, for the purposes in this policy only.

8. Retention, and what happens when your agreement ends

During your agreement, we retain your machine data (sampled images, grade results, and telemetry) to support your machine and track model performance over time.

When your agreement ends, we stop collecting data from your operation. Data already collected stays in our systems. It is used only to improve the AI and to publish results under Section 6, and Sections 5 and 6 continue to apply to it permanently. The same applies to data from a demonstration if you do not go ahead with an agreement.

9. Changes to this policy

When we change this policy, we update the date and version at the top and publish the new version at twistertrimmer.com/marvel/data-policy. If a change materially affects how your data is used, we will notify you before it takes effect. No change reduces the protections in Sections 5 and 6 for data collected before the change takes effect, and no change overrides your signed agreement.

10. Contact

Questions and requests about machine data: privacy@keirton.com
Privacy Officer, Keirton Inc.