Keepso Privacy Policy
Effective: November 6, 2026 · Operator: Hiorio (HWIHO KIM)
Information stored on your device
Selected photos, links, notes, bundles and saved summaries are stored on your device. AI summaries and automatic link classification use data only after consent to the transfer for each feature. Deleting the app may delete your on-device library. Export important content from Settings.
Importing Instagram posts
Before the first external import, you receive a separate consent request for sending the link to Cloudflare and Apify and an explanation of content-processing permission. Until you consent, links are saved on your device and external post collection does not start. When you request an import of a public Instagram post link, the link is sent through our Cloudflare server to Apify to retrieve public information such as the caption, author, posting time and photo URLs. We do not request access to private posts or your Instagram password. Imported photos, captions and original sources are stored in your on-device library. To prevent duplicate execution and allow retries, the server temporarily retains post information and processing status for up to 24 hours, then cleans it up on subsequent access. Apify processes requests and collected results on separate infrastructure; retention follows the Apify Privacy Policy and Data Processing Addendum. AI analysis during link import requires separate consent to automatic classification. Post-import consent is separate from AI photo and caption transfer consent. Withdraw consent for future transfers in Keepso Settings; saved links and content remain.
AI processing of selected photos
When you request an AI summary and consent to the transfer, we send the selected photos, any saved post caption, and summarization instructions through our Cloudflare server to the Google Gemini API or, if you expressly consent to the OpenAI transfer, to the OpenAI API. When available, summaries in the regular release use a connected paid API. Text and personal information in the photos may also be processed, so exclude sensitive photos. The purpose is to generate your requested summary; the production server does not store photos as separate files. To recover results after network errors, generated summaries are temporarily available for up to 10 minutes and are cleaned up on subsequent access. Transfers may be processed on Cloudflare, Google, or OpenAI infrastructure outside your country. Google's processing and retention follow the Gemini API Terms. We turn off response storage for OpenAI requests. By default, OpenAI may retain inputs and outputs in abuse-monitoring logs for up to 30 days, subject to exceptions. See the OpenAI API data controls. You can withdraw consent for future AI transfers in Settings.
Instagram video summaries
When you request a video summary for a public Instagram post and separately consent to sending the video, our Cloudflare server retrieves the original video and sends its visuals and audio to the Google Gemini API along with any photos and captions included in the summary request. Video summaries in the regular release use a connected paid Gemini API. The TestFlight app can use the free Gemini API. Keepso does not retain the original video as a separate file. You can withdraw consent for future video transfers in Settings.
Trial, purchases and balance verification
After verifying Apple’s signatures, we store decoded app transaction identifiers, original and individual purchase transaction identifiers, account-linking tokens, products, store environments, purchase dates, subscription periods and status, and credit grant, usage and refund records on the server. The identifier and token values themselves are included in these purchase records. This prevents duplicate initial 6-credit trials and duplicate purchase or usage delivery for the same App Store account, and restores the accumulated balance. The original signed app and purchase transaction payloads (JWS) used for verification are not stored. Apple does not provide us with your name, email or card number. Purchase, grant, usage and refund verification records, request identifiers, input hashes, processing status and amounts charged are retained for restoration after reinstalling and to prevent duplicate processing. Temporary summaries are deleted on access after up to 10 minutes.
Contact and exercising your rights
Inquiry emails and their contents are used to reply. You can request access to or deletion of personal information at support@hiorio.com. App transaction identification information may be needed to identify you. We delete information on request except for transaction records required by law.
External services
Apple handles purchases and payments, Cloudflare handles the API and usage storage, Google and OpenAI process requested AI summaries, and Apify handles requested public Instagram post imports. We do not use advertising SDKs or user tracking. If this policy changes, the effective date and changes will be posted on this page.
TestFlight test app
Apple-verified TestFlight installations can test Instagram imports and AI summaries regardless of the credit balance, without deducting credits. The regular release uses the initial 6-credit trial and 100-credit subscription. Older APIs are blocked with an update notice. The free Gemini project's maximum of 500 requests per day is shared by all test installations and resets at midnight Pacific Time. Provider per-minute and token limits may temporarily restrict requests. Use only material without personal, confidential, or sensitive information in these tests. After the free daily limit is exhausted, we may use the paid OpenAI API if connected and you have consented to transfer to OpenAI. A separately connected paid Gemini API may also be used. Additional requests stop when no eligible paid API is available. This switch does not charge test users through the App Store. The regular release also offers a separate 100-credit top-up; all credits accumulate and never expire.
Optional automatic link classification
If you separately consent to automatic link classification, the link’s text, any text read from photos on your device, existing collection names and short examples from saved items are sent through Cloudflare to Google Gemini or OpenAI to choose a matching collection. For ordinary web links, your device visits the source page to read its text and saves the extracted text in your local library. This classification uses no additional wallet credits. The server keeps only the request identifier, input hash, status and selected collection result for up to 48 hours, then removes them on subsequent access. Classification input is not stored as a separate server file. The provider processing and retention conditions described above also apply. You can withdraw this consent in Settings.
Free Gemini API in the test app
Inputs and generated results sent to the free API may be used to improve Google's services or reviewed by people. Do not send videos, photos or captions containing personal, confidential, or sensitive information. See the free services section of the Gemini API terms for detailed processing conditions.