Rivetra AI and Automation Disclosure

Policy version: 2026-07-12

Effective date: July 12, 2026

Rivetra uses artificial intelligence and rules-based automation to assist contractor workflows. This Disclosure explains capabilities, limitations, required human review, data use, and prohibited conduct.

1. AI Features

Depending on plan and availability, Rivetra may use AI to:

• draft estimates, line items, scopes, assumptions, questions, and job audits;
• extract receipt, supplier quote, and document information;
• research permit offices, codes, and jurisdiction resources;
• draft contracts, change orders, messages, and project summaries;
• classify financial records and suggest accounting actions;
• draft website, SEO, GEO, AEO, social, and review-request content;
• summarize projects, conversations, reports, and business activity; and
• recommend next steps or prepare user-approved actions.

2. AI Output Is Not a Fact or Final Decision

AI can produce false, incomplete, outdated, inconsistent, biased, or misleading output. A confident tone does not mean an output is verified. Rivetra may display confidence, source, catalog, model, and audit information, but those indicators do not guarantee correctness.

Users must independently review material output before relying on it. Rivetra does not provide legal, tax, accounting, engineering, architectural, environmental, employment, licensing, code, permit, safety, or financial advice.

3. Human Approval Gate

Unless a feature is clearly designated as a low-risk, user-configured automation, human approval is required before AI output:

• changes an estimate, invoice, contract, change order, ledger, payment,

refund, tax record, permit record, or company setting;

• is sent to a customer, employee, subcontractor, vendor, or regulator;
• is published to a website, search profile, Facebook, Instagram, or another

public channel;

• deletes, archives, or materially changes a record; or
• represents a legal, safety, licensing, inspection, code, or permit conclusion.

Approval means the user reviewed the actual output, not merely that an AI task was requested.

4. Estimating and Financial Safeguards

AI-generated quantity, price, labor, tax, margin, discount, and cost data are drafts. Users must confirm measurements, supplier prices, labor burden, availability, taxability, waste, overhead, profit, exclusions, and customer requirements.

Financial actions use deterministic cents-based calculations after structured validation. AI may suggest a classification but may not silently post or alter money records. Rivetra is not a substitute for a bookkeeper, CPA, tax adviser, payroll provider, or financial institution.

5. Permit, Code, and Local Research

Permit and code information must be tied to the jobsite ZIP, city, county, state, and authority having jurisdiction when available. Rivetra should favor official government sources and distinguish a verified source from an AI inference.

Users must open and verify official sources before treating a permit, fee, inspection, code, zoning, licensing, or procedure statement as current. Rivetra cannot determine that a permit is approved, unnecessary, or sufficient for a specific job. A permit allowance is not a government fee quote or approval.

6. Marketing, Reviews, and Publishing

AI may draft content only from authorized business and job facts. Users may not use Rivetra to create or publish:

• fake or false reviews, testimonials, ratings, endorsements, awards, licenses,

customer quotes, before-and-after claims, or completed-job claims;

• reviews by a person who did not have the represented experience;
• incentives conditioned on positive sentiment;
• undisclosed insider, employee, owner, family, or paid endorsements;
• deceptive location, service-area, credential, availability, or pricing claims;

or

• content that suppresses, threatens, or misrepresents genuine negative reviews.

Public content requires a named user approval record. Provider permissions do not authorize Rivetra to publish unapproved content.

7. Prohibited Uses

Users may not use AI or automation to:

• facilitate fraud, impersonation, forged signatures, false documents,

discrimination, harassment, malware, credential theft, or unlawful activity;

• infer or make consequential decisions based on protected characteristics;
• create unsafe construction, environmental, electrical, plumbing, structural,

lead, asbestos, mold, or specialty-work instructions without qualified review;

• evade safety controls, rate limits, subscriptions, provider rules, or audit

logs;

• submit secrets, full payment credentials, password hashes, or usable OAuth

tokens; or

• represent AI output as government, professional, or Rivetra certification.

8. Data Used for AI

When a user requests an AI feature, Rivetra may send relevant prompts, company settings, project details, measurements, images, receipts, documents, and prior approved context to the configured AI provider. Rivetra seeks to minimize the data and exclude secrets and unrelated personal data.

Do not submit unnecessary sensitive personal data, children's data, biometric identifiers, government identification, health data, or privileged legal communications. Provider processing and retention depend on the applicable Rivetra provider configuration and contract.

9. Audit and Traceability

Rivetra may record the requesting user, company, project, feature, prompt type, model and version, structured response, source links, confidence information, approval or rejection, final action, and timestamps. These records support quality review, security, dispute resolution, and compliance.

Model or prompt changes may cause different outputs from the same input. Cached results may be returned when the relevant job facts and evidence have not changed.

10. Automation

Low-risk automation must be opt-in, scope-limited, permission-controlled, observable, rate-limited, and reversible where practical. Integration errors must produce a human-readable recovery state rather than silently retrying a consequential action.

Rivetra may suspend an automated action when confidence is low, a provider is unavailable, a source cannot be verified, approval is missing, or abuse is suspected.

11. Reporting AI Problems

Users should report unsafe, deceptive, inaccurate, privacy-invasive, or unexpected AI output through the in-app feedback function or support@rivetra.app. Security vulnerabilities should be sent to security@rivetra.app. Include the job or feature and enough context to investigate, but do not email secrets or full payment credentials.

12. Changes

Rivetra will update this Disclosure as models, features, providers, and laws change. Material changes will receive additional notice or renewed acceptance when required.