PurVago — AI Features Disclosure & Responsible-Use Terms

Effective date: October 18, 2026 for accounts opened before September 18, 2026; on acceptance for accounts opened since · Last updated: September 23, 2026 · Previous versions

PurVago includes features powered by artificial intelligence and machine learning ("AI Features", as defined in Section 1.7 of the Terms of Service). This disclosure explains what they do, their limitations, how data is handled, and your responsibilities. It is incorporated into and governed by the Terms of Service (https://app.purvago.com/legal/terms), including its dispute-resolution and limitation-of-liability provisions.

1. What the AI Features are

Depending on your plan and configuration, AI Features may include:

• Vision estimating — analyzing photos and video frames of a property (from office staff, field staff, or a homeowner's video-estimate link) to help generate estimate line items, quantities, or suggested pricing.
• Walkthrough transcription & analysis — transcribing the voice notes recorded during a property walkthrough and using them in the analysis.
• Note polishing — rewriting walkthrough notes into brief, client-ready wording.
• Academy course drafting — PurVago staff can draft training-course text and images with AI. Customers and their teams do not use this tool, and quizzes are scored against fixed answer keys, not by AI. Assessments record training completion for the Customer's own team. PurVago does not use them to make, and they are not designed to make, hiring, promotion, or retention decisions.
• Other assistive or generative features we may add, which will follow Section 2A.

Not AI Features: tap-to-dictate converts speech to text with your device's or browser's own speech recognition (for example Apple's on iPhone); PurVago does not receive the audio. Suggested time slots and route ordering are calculated by fixed rules, without an AI provider.

2. How they work (and who processes the data)

AI Features process the inputs you provide (e.g., images, audio, text) to produce outputs. Processing occurs through our AI model subprocessor (OpenAI, through its API) and through our hosting platform; see the Subprocessors list (https://app.purvago.com/legal/subprocessors). Inputs and outputs are handled in accordance with the Privacy Policy and the Data Processing Addendum. Voice inputs are transcribed to text; we do not create voiceprints, perform speaker identification, or otherwise use voice for biometric identification. OpenAI processes inputs only to return results to PurVago, does not use them to train its models, and may keep them for up to 30 days to detect abuse before deleting them.

2A. Permission before anything is sent

• Each person is asked. Before any photo, video frame, recording, transcript or note is sent to OpenAI, the Service asks the signed-in person for permission, once. If they decline, nothing is sent and the AI step is skipped; they can do the same work by hand, and they are asked again the next time they use an AI Feature.
• Withdrawing. Anyone can withdraw or give permission at any time under the user menu > Privacy & AI. After withdrawal nothing more is sent.
• Homeowners. A homeowner sending photos through a video-estimate link chooses with an unticked box whether AI may read them. Without it, no AI runs and the cleaning company reviews the photos itself.
• Your separate duty. A team member's permission covers sending their captures to OpenAI. It does not replace the Customer's own responsibility, under Section 4, to have the property owner's or occupant's consent to photograph and record.

3. Outputs are assistive, not authoritative — verify before you rely

3.1 No guarantee of accuracy. AI outputs — including estimates, quantities, prices, cost/margin figures, transcriptions, summaries, and property information — are probabilistic and may be incomplete, inaccurate, outdated, or wrong. They can "hallucinate" details that are not real. 3.2 Human review required. AI Features are decision-support tools. You are responsible for independently reviewing and confirming every output before relying on it, quoting it to a client, issuing an invoice, scheduling work, making a payment decision, or taking any other action. Do not treat AI output as a final quote or as fact without verification. 3.3 Not professional advice. AI Features do not provide legal, tax, accounting, insurance, safety, employment, or other professional advice, and are not a substitute for professional judgment or a physical inspection. 3.4 Estimates are not binding quotes. Any AI-assisted estimate is an internal aid; the price you commit to your client is set by you.

4. Your responsibilities and consents

4.1 Rights & consent to inputs. You represent that you have all necessary rights and permissions to submit your inputs to the AI Features, including the property owner's/occupant's consent to photograph the premises and to record audio/video. 4.2 Recording-consent laws. Walkthrough recording and dictation may capture the voices of individuals. Recording laws differ by jurisdiction, and some require all-party (two-party) consent. You are solely responsible for obtaining any legally required notice and consent before recording, and for complying with applicable wiretap/eavesdropping, biometric, and privacy laws. 4.3 Prohibited inputs. Do not submit content you are not authorized to use, content that infringes third-party rights, or unlawful, harmful, or highly sensitive personal data beyond what the feature is designed to handle. Do not use AI Features to make solely automated decisions with legal or similarly significant effects on individuals without appropriate safeguards and human oversight. 4.4 Employment decisions. If you use any AI Feature output in connection with a hiring, promotion, or retention decision, you are the deployer of that tool for the purposes of laws that regulate automated employment decisions (for example, New York City Local Law 144 and the Colorado Artificial Intelligence Act), and you are responsible for any bias audit, notice, and record-keeping those laws require. Tell us at support@purvago.com if you intend such use so we can provide the information about the feature that those laws require of a developer. 4.5 No reverse engineering / model abuse. Do not use AI Features to extract or scrape model weights or training data, or to circumvent safety or rate limits.

5. Use of data for model improvement

5.1 We use inputs and outputs to provide and operate the AI Features. We may use data that has been de-identified so that it does not identify you or any individual to improve the Service. We will not attempt to re-identify such data, will maintain it in de-identified form, and will contractually obligate any recipient to the same. 5.2 We contract with our AI model providers on terms that prohibit the use of your inputs and outputs to train their models. We will maintain such terms with any AI provider we engage, and will update the Subprocessors list in accordance with Section 6.3 of the DPA.

6. Bias, fairness & limitations

AI systems can reflect biases or produce inconsistent results across inputs and conditions (e.g., image quality, lighting, accents, or edge cases). Outputs should not be used as the sole basis for decisions affecting individuals' rights or employment. You are responsible for using AI Features in a non-discriminatory, lawful manner.

7. Availability & changes

AI Features depend on third-party providers and may change, degrade, or become unavailable, and may be modified, limited, or discontinued. Usage may be rate-limited or metered.

8. Disclaimer & liability

8.1 AI Features are provided "AS IS" and "AS AVAILABLE," without warranties of any kind, as further described in Section 11 of the Terms of Service. 8.2 To the maximum extent permitted by law, PurVago is not liable for any decision made, action taken, quote given, amount billed, or loss incurred in reliance on AI outputs. The limitations and cap in Section 13 of the Terms of Service apply to AI Features. You indemnify PurVago for claims arising from your use of AI outputs, per Section 14 of the Terms.

9. Reporting problems

Report inaccurate, unsafe, or concerning AI output to support@purvago.com so we can investigate and improve the Service.