Privacy Policy

Last updated: May 2, 2026

TextAgent ("we", "our", "us") is committed to protecting your privacy. This Privacy Policy explains how our iOS application handles your information.

1. Core Privacy Principle

TextAgent is designed with a privacy-first architecture. SMS classification and filtering runs entirely on your device by default; no SMS content is automatically uploaded for filtering purposes. SMS content only leaves your device in two explicit opt-in scenarios described in Section 5 and Section 6: (a) when you choose to share specific messages to help improve our classification model, or (b) when you actively type or paste text into the AI Assistant or Agent chat.

2. How SMS Filtering Works

TextAgent uses Apple's IdentityLookup framework to filter unknown sender messages. This framework is designed by Apple with strict privacy constraints:

• The Message Filter Extension runs in a sandboxed environment on your device
• It can only classify messages as Allow, Junk, or Transaction subcategories
• It cannot access your contacts, photos, or other personal data
• Message content is processed locally during filtering and is not automatically sent to our servers

3. Data We Collect

We collect minimal data necessary to provide our services:

• Account information: If you create an account, we store your email address for authentication
• Subscription status: Your subscription tier (Free, Pro, Ultra) for feature access
• Rule metadata: Filter rules you create or subscribe to (rule patterns only, never message content)
• Usage statistics: Aggregate counts of AI chat and Agent feature usage for subscription quota management
• AI conversation content: When you interact with our AI Assistant or Agent features, we record your messages and the AI's responses on our servers for service quality assurance, abuse detection, and usage auditing (see Section 6 for details)
• User-submitted training samples (opt-in): If — and only if — you explicitly tap "Share this message to improve classification" or submit feedback on a mis-classified message, the specific SMS content you selected is uploaded to our servers to improve future versions of our on-device Core ML classification model. This is always initiated by you on a per-message basis; we do not bulk-collect or upload messages without your explicit action
• Crash reports: Anonymous crash data to improve app stability

4. Data We Do Not Collect Automatically

• SMS or text message content (not collected automatically; only the two opt-in scenarios described in Sections 3 and 6 can cause a message to leave your device)
• Phone numbers from your messages
• Your contacts or address book
• Your location data
• Browsing history or app usage from other apps

5. On-Device Processing

TextAgent uses multiple classification layers, all running on your device:

• Rule Engine: Pattern matching against your configured filter rules
• CoreML Model: A machine learning model that runs entirely on-device
• AI Classification: When enabled, uses API calls for complex classification — only the classification result is returned, and message content is processed in-memory without persistent storage

6. AI Conversation Data

When you use the AI Assistant or Agent features (available to Pro and Ultra subscribers), your interactions are processed through our backend servers. We collect and store the following:

• Your messages sent to the AI Assistant or Agent
• The AI-generated responses returned to you
• Tool calls and their results during Agent task execution
• Token usage and request metadata (model name, latency, timestamps)

This data is used exclusively for:

• Service quality monitoring and improvement
• Abuse and misuse detection
• Subscription usage quota enforcement
• Troubleshooting technical issues

Important clarifications:

• This section only applies to content you actively send in AI Assistant and Agent conversations
• Automatic SMS filtering and classification always happens on-device; filter results are never sent to our servers or to AI providers
• If you paste an SMS into an AI chat yourself, that pasted text follows the AI conversation data flow described above — because you chose to send it
• AI conversation logs are automatically deleted after 90 days
• You can request deletion of your conversation data at any time by contacting us
• We do not use your conversation data for advertising or share it with third parties

7. Rule Marketplace

When you use the Rule Marketplace:

• We store rule definitions (patterns and categories) on our servers
• Subscription counts and ratings are aggregated anonymously
• No message content is ever shared through the marketplace

8. Third-Party Services

TextAgent uses the following third-party services:

• Apple App Store: For subscription management and payments
• Alibaba Cloud (Qwen / 通义千问): Our primary AI service provider. When you use AI Assistant or Agent features, your conversation messages are transmitted through TextAgent's backend servers to Alibaba Cloud DashScope API (Qwen model). Data is transmitted over HTTPS and processed solely for generating AI responses. Automatic SMS filtering never sends SMS content to AI services; SMS content reaches an AI provider only if you explicitly paste it into an AI chat yourself.
• Additional AI providers: We may integrate other AI services in the future, including but not limited to OpenAI (ChatGPT), Anthropic (Claude), and Google (Gemini). You will be notified and asked for renewed consent before any new AI provider is used to process your data.
• Supabase: For user authentication and account management

Data Flow: Your AI chat messages → TextAgent backend servers (encrypted HTTPS) → AI service provider API → AI-generated response → TextAgent backend → your device. We do not store your chat messages indefinitely — conversation logs are automatically deleted after 90 days.

In-App Consent: You must explicitly agree to AI data sharing during onboarding (Step 3 of 5) before any data is sent to AI services. The consent screen clearly identifies every third-party AI provider, the exact data transmitted, and what is never sent. You can revoke this consent at any time in the app's Settings > Privacy & Security. Upon revocation, AI features will be disabled and no further data will be transmitted.

Use of Your Data by AI Providers — What We Can and Cannot Guarantee: We access AI services through their standard APIs; we therefore rely on, and are bound by, the public API terms each provider publishes at the time of the request. As of the date of this policy, the default API terms of the providers we integrate state the following (please verify on each provider's website, as terms may change):

• OpenAI API — Per OpenAI's API data usage policy (effective March 1, 2023), data submitted via the API is not used to train OpenAI models by default, and is retained for up to 30 days for abuse and misuse monitoring before deletion.
• Anthropic Claude API — Per Anthropic's Commercial Terms, inputs and outputs submitted via the API are not used to train Anthropic's models by default.
• Google Gemini API — Per Google's Gemini API Additional Terms, data submitted on the paid tier is not used to improve Google's products; data submitted on the free tier may be used by Google to improve its services. TextAgent uses the paid tier for production traffic.
• Alibaba Cloud DashScope (Qwen) — Per Alibaba Cloud's DashScope service terms, API requests may be retained for a limited period for safety review and service quality purposes. Alibaba Cloud does not publicly state that paid API inputs are used for model training, but we cannot independently verify the internal handling.

What this means honestly: we do not have private negotiated contracts with every provider that override their public terms, and we cannot independently audit what each provider does inside their infrastructure. What we do control, we disclose plainly:

• We transmit only what is necessary to generate the AI response — your chat message plus limited system instructions.
• We do not sell, rent, or share your data with advertisers or data brokers.
• TextAgent's own backend logs related to AI requests are retained for up to 90 days solely for abuse detection, quota enforcement, and service quality monitoring, and are never used for AI model training.
• If a provider changes its terms in a way that materially expands data use (for example, starting to train on API data by default), we will update this policy and, where required, seek renewed consent inside the app before continuing to route your data to that provider.

Each Provider Has Its Own Privacy Policy: Because we cannot guarantee provider-internal behavior, we strongly encourage you to review each provider's own terms and privacy policy directly — Alibaba Cloud, OpenAI, Anthropic, and Google — before using AI features. If you are not comfortable with a provider's terms, please do not grant AI data consent or, if already granted, revoke it in Settings > Privacy & Security.

9. Data Storage and Security

• Local data is stored in your device's secure storage using iOS App Groups
• Sensitive data (API keys) is stored in the iOS Keychain
• Server communications use HTTPS encryption
• We do not sell your data to third parties

10. Your Rights

You have the right to:

• Delete your account and all associated data
• Export your filter rules
• Opt out of analytics collection
• Request information about data we hold about you

11. Children's Privacy

TextAgent is not directed at children under 13. We do not knowingly collect personal information from children.

12. Changes to This Policy

We may update this Privacy Policy from time to time. We will notify you of significant changes through the app or via email.

13. Contact Us

If you have questions about this Privacy Policy, contact us at hello@textagent.net.