AI Usage Disclosure

Rishtio is an AI-powered matrimony service. AI supports profile writing review, search, matching, recommendations, and relationship-manager assistance. Users and families remain the decision-makers. This disclosure identifies the current provider, data, purposes, stored outputs, and controls for the United States launch.

Microsoft Azure OpenAI in East US processes natural-language search text, the active saved preference, and relevant profile context used to understand relative requests such as "same state" or "younger than me." Context may include age derived from date of birth, city/state/country, religion, community/caste, languages, diet, education, income currency, smoking and drinking habits, and the current profile summary. The service returns structured filters and may generate a short saved-search name.

Azure OpenAI generates numerical profile and preference embeddings used for semantic ranking. Profile embedding input currently may include age, religion, community/caste, city/state/country and hometown, nationality, education, job and company, diet, marital status, smoking and drinking habits, About Me, hobbies, personality traits, and languages. Preference embedding input includes the saved ideal-partner description. Rishtio stores the resulting profile vector and semantic scores. We do not send identity documents, face-verification captures, chat messages, or contact details to Azure OpenAI for these embeddings.

Azure OpenAI may generate a multilingual profile summary from profile details including name, age, height, location and hometown, religion/community, education, job/company, income, diet, marital status, hobbies, About Me, and personality traits. It may generate profile-based search suggestions from coarse location, education, work, diet, relocation, languages, interests, personality, and core values. For writing review, it receives the exact current narrative draft, interface language, field, and authorship mode. The reviewed text is published only after the model returns READY and the user separately finalizes that exact text. Profile summaries and search suggestions are stored until regenerated or deleted. Review receipts expire and cannot be reused for changed text.

Rishtio Select uses pinned saved preferences, profile embeddings, deterministic filters, and stored semantic scores to surface candidates for review. Current launch scoring does not apply free-form custom ranking instructions and does not generate discovery-conversation summaries. Select does not contact people or make relationship decisions autonomously. Turning Select off leaves manual search and matchmaking available.

AI outputs are suggestions. You decide what profile text to save, which filters to use, whether to enable Select, and whom to contact. Rishtio does not promise universal human review of each generated output. Users can flag offensive profile summaries, search suggestions, or writing reviews in the app for moderation review. For profile-photo uploads, configured Content Safety severity signals can block an upload but do not decide relationship or account suitability; support is the dispute path. Required identity document verification is a narrow signal about the configured document check. Face verification is unavailable. Neither moderation nor verification is a background, character, intent, truthfulness, compatibility, or safety guarantee.

Use the in-app flag on a generated summary, suggestion, or writing review to report offensive output without placing the generated text in telemetry. Dispute an upload-safety result by emailing ai@rishtio.com, or by creating a support ticket if you sign in to your own Rishtio account. Include enough context to locate the event, but do not email passwords, tokens, identity documents, or biometric captures. For privacy requests or an appeal of a denied request, contact privacy@rishtio.com, or create a support ticket from Settings → Privacy & Support if you sign in to your own Rishtio account. We respond within the period required by applicable law and do not promise a fixed response time that is not operationally enforced. Rishtio does not use AI to make solely automated decisions that produce legal or similarly significant effects. Where applicable state law provides rights related to qualifying profiling, you may exercise them through the Privacy Team.

The profile subject must accept the current account agreement before using Rishtio. The agreement covers essential AI processing described in the Terms and Privacy Notice. An assistant cannot accept it for a managed profile, and an unclaimed profile is not processed by these AI features. Rishtio does not offer a separate non-AI mode. You control which optional profile fields you supply and can edit or delete those values. You may also delete your account from Settings → Account actions → Delete Account. A materially new provider, data category, or purpose will be reflected in updated legal documents and handled as required by applicable law.

Data is encrypted in transit. The current Azure OpenAI deployment processes prompts and responses in East US. Microsoft states that customer prompts, completions, embeddings, and training data are not available to OpenAI or other customers and are not used to train or improve Microsoft or third-party models without explicit permission or instruction. The inference models are stateless. Content flagged for potential abuse may be sampled for automated or authorized human review under Microsoft's abuse-monitoring terms unless modified monitoring has been approved. For ordinary profile-photo uploads, Azure AI Content Safety receives the whole image in East US and returns Hate, Sexual, Violence, and Self-Harm severity signals under the configured thresholds. Assistant photos, identity documents, Share Kit files, and support uploads are never sent. Azure AI Document Intelligence separately processes identity documents when the profile subject submits one under the account agreement. We minimize provider inputs and do not log image content, verification documents, or provider response bodies. Rishtio stores the derived AI outputs described above in Central US until replacement or account deletion.

AI systems may misinterpret text, omit context, generate inaccurate or offensive wording, or reflect statistical bias. Semantic similarity and Select scores do not measure character, safety, compatibility, or the likelihood of a successful relationship. Rishtio does not guarantee the accuracy, completeness, fairness, or reliability of generated output.

For questions about our AI practices or to exercise your rights regarding AI processing, contact us at: Email: ai@rishtio.com