Common Analytics Limited · Updated 22 October 2026
AI Discovery privacy notice
This notice covers the AI Discovery workspace. It explains the pilot’s actual data handling and its limits.
What we collect and why
We collect your email and company website, your project answers and estimates, shared comments, conversation, text examples, invitation details and discovery decisions. We use these to guide one workflow discovery, save and share the project, produce a brief, review the opportunity and respond to your requested discussion or privacy request. We keep a versioned record of the notice acknowledged at project creation.
We also use session identifiers, expiring sign-in challenges, request-limit records and project activity history to operate and protect the workspace. Request limits use hashes of email or IP values; a hash is not a guarantee of anonymity.
Who can see project information
Every accepted member of a project can see its shared answers, examples, files, comments and brief. Inviting someone grants access to the whole shared workspace. Authorised Common Analytics staff can review projects and contact participants about them. Internal staff assessments are excluded from participant views, participant exports and AI prompts.
A company email domain does not grant access to other projects. Entering an email without verifying it creates only a browser-owned guest project. Verify later to secure access across devices. We cannot recover an unclaimed guest project merely because someone enters the same email.
Services used to process the information
Vercel runs the application, Neon Postgres stores project records, private Vercel Blob stores uploaded text originals, Microsoft Azure OpenAI processes excerpts for AI guidance, and Amazon SES delivers requested sign-in codes. Neon and the configured private Blob store are in Singapore. Azure’s processing location and provider operational retention depend on the resource/deployment configuration; we do not promise that every service processes data only in Hong Kong or Singapore.
Azure requests contain bounded excerpts of the current project’s shared answers, evidence and conversation. They exclude account email fields and internal assessments, but personal information entered into shared text may still be included. We request no stored Responses history (store: false) and give the model no tools. This setting does not establish zero provider logging, zero abuse-monitoring retention or immediate physical erasure. Providers may retain operational, security, delivery or recovery data under their own terms and settings.
What you should leave out
Share only information you are permitted to provide. Redact personal and customer identifiers where possible. Do not submit passwords, API keys, financial account details, health records or sensitive personal information. Text is not automatically anonymised. This pilot accepts UTF-8 text examples only; PDF and image uploads are not supported.
Retention and scheduled deletion
The free discovery window normally lasts seven days. Project information is scheduled for deletion 90 days after that window ends, including any explicitly agreed extension. Editing answers, asking for a discussion or signing in does not restart retention. The workspace shows the scheduled date.
Access is blocked when that deadline is reached. Scheduled maintenance removes the project records, history, comments, invitations, staff assessments and files. Cleanup can be delayed if maintenance or a provider is unavailable; access stays blocked while deletion is retried. New hosted discovery intake closes if the last successful maintenance check is more than 24 hours old.
Sign-in codes expire after ten minutes; sessions expire after seven days. Expired challenge, session and request-limit records are cleared by maintenance. Privacy support requests are kept for up to 90 days while open or 30 days after closure. Minimal deletion receipts contain random project/receipt references, dates, reason category and deletion state; they are kept for 180 days to prevent accidental restoration. They do not retain project content or participant email addresses.
Export, correction and deletion
Use “Privacy, export and deletion” inside the project to download shared project data or save an access, correction or deletion request for staff. These requests do not send an automatic email. You can also contact john.l@common-analytics.com. We verify identity and the relevant project access before disclosing or changing information.
The project owner or authorised staff can delete the whole project by confirming its current name. All members immediately lose access. Records are removed from the active store, and private-file removal is attempted immediately with retries if needed. A contributor can request help with their own information; they cannot independently erase colleagues’ shared project. A deletion receipt distinguishes completed private-file cleanup from a pending retry.
Removing a single evidence item invalidates the brief but may leave participant answers, quoted statements or earlier derived claims that refer to it. Ask staff to review those references, or use whole-project deletion when the entire project should be erased.
Backups and deletion limits
Application-managed backups are encrypted and retained for 7 days. Scheduled maintenance removes deleted projects from existing application backups as well as from future backups. Archives older than 7 days are rejected by the recovery tool and removed at the next successful maintenance run. Backup cleanup may therefore be delayed while maintenance is unavailable. Recovery checks the independent deletion journal and keeps recovered projects isolated with sharing locked.
Neon’s provider-managed recovery history is separate from these application backups. The current project’s configured recovery window has not been verified with the available provider access; we do not promise a shorter provider erasure deadline. Deleted versions may remain in restricted provider recovery copies until those copies expire. A provider restore must reapply the deletion journal before any project is reopened. If the journal’s completeness cannot be established, recovery stays offline. Common Analytics cannot recall files or briefs already downloaded or copied by participants.
AI output and contact
AI can make mistakes. Discovery outputs are preliminary hypotheses and participant estimates, not a binding scope, quote, professional advice or guaranteed saving. Starting discovery does not subscribe you to marketing emails, book a meeting or commit you to a purchase.
Privacy contact: john.l@common-analytics.com
Common Analytics Limited
Unit 673, 6/F, Building 19W, No. 19 Science Park West Avenue, Hong Kong Science Park, Pak Shek Kok, N.T.
Policy version: 2026-10-22.privacy.v1