DRAFT — not legal advice — lawyer must review before launch
Cofello Privacy Policy
Version: v1.1 draft (recommended defaults added Oct 18, 2026 — see decisions-v1.md) · Effective date: [EFFECTIVE DATE] · Last updated: [LAST UPDATED DATE]
<!-- Drafting key used in every Cofello draft: "REQUIRED" = clause driven by a law or by a platform contract Cofello must follow (Google, Slack, Stripe). "GOOD PRACTICE" = recommended, not strictly required. Each note names its source. -->
1. Who we are and what this policy covers
Cofello is a workspace where you direct a team of AI agents that help run your business. The service is run by [COMPANY LEGAL NAME], a Delaware corporation ("Cofello", "we", "us"), at [REGISTERED BUSINESS ADDRESS]. Our websites include cofello.ai and joincofello.com.
This policy explains what personal data we collect, why, who we share it with, how long we keep it, and the choices and rights you have. It covers:
- people who visit our websites;
- people who sign up for and use Cofello ("users"), including workspace owners and invited seats;
- marketplace creators and buyers; and
- people whose data reaches Cofello because a user connected an account, such as the senders and recipients of emails in a connected Gmail account or members of a connected Slack workspace.
Privacy contact: [PRIVACY EMAIL]. Data protection officer (if appointed): [DPO NAME AND CONTACT, OR "NOT APPOINTED"].
<!-- REQUIRED | GDPR Art. 13(1)(a)-(b): identity and contact details of the controller and any DPO | https://gdpr-info.eu/art-13-gdpr/ -->
<!-- REQUIRED | Google requires a public privacy policy listed in the OAuth client configuration | https://developers.google.com/terms/api-services-user-data-policy -->
2. The short version
- We do not sell your personal data. We do not use it for advertising.
- We do not use your Gmail or Slack data, or any customer content, to train or improve AI models, ours or anyone else's. We only use the AI-provider options whose terms say inputs are not used for training (see section 8).
- Your connected data stays in your business workspace. It is never used to benefit another customer, never feeds marketplace rankings or results, is never shown to marketplace creators, and never goes into a clone that is listed or sold. [DECIDED by Chase, Oct 18, 2026]
- By default your agents need your approval before they send an email. You can change that setting for your own account.
- You can delete your data, disconnect accounts at any time, and remove seats. Section 11 explains what is deleted and when.
3. Our two roles: processor and controller
Data protection law gives different duties depending on who decides how data is used.
| Data | Our role | What that means |
|---|---|---|
| **Customer content**: data in connected accounts (Gmail, Slack and other tools), files you upload, chat with agents, your business memory vault, agent outputs, clone training material you provide | **Processor** (in the US, "service provider") acting for the customer | The customer (the business that owns the workspace) decides what is connected and why. We process it only to provide Cofello to that customer, under our Data Processing Addendum (`dpa.md`). |
| **Account and billing data**, website and product usage data, support messages, marketplace creator and payout data, security logs | **Controller** | We decide how this data is used, and this policy applies directly. |
If you are someone whose email or Slack message reached a Cofello customer's workspace, the customer is responsible for that data. Please contact them first. We will help them answer you and will pass your request on if you send it to us.
<!-- REQUIRED | GDPR Arts. 4(7), 4(8), 28 and EDPB Guidelines 07/2020 on controller/processor | https://www.edpb.europa.eu/our-work-tools/our-documents/guidelines/guidelines-072020-concepts-controller-and-processor-gdpr_en -->
<!-- LAWYER: confirm the split. Launch-blockers row 7 notes that any cross-tenant learning would likely make Cofello a controller of that data; this draft avoids that by banning cross-tenant use (section 7). -->
4. What we collect
Data you give us
- Account data: name, email address, password (stored hashed) or sign-in provider identifier, business name, role, the seats you invite.
- Onboarding answers: your goals, which tools you use, how much access and spending authority you give your agents.
- Billing data: billing name and address, tax ID if you give one, plan and credit purchases. Card details are collected and stored by Stripe, not by us. We receive a token, card brand, last four digits and expiry.
- Content you add: chats with agents, files, documents, recordings and notes you upload, memory entries, corrections, approvals and rejections.
- Clone material: if you build a clone of yourself, the recordings, documents and interview answers you choose to provide, and your consent record (section 9).
- Marketplace data (creators): listing content, prices, identity and payout details collected through Stripe Connect, tax forms.
- Support and communications: what you send us, and survey answers.
Data from accounts you connect
- Gmail / Google: see section 6 for exactly what we access and why.
- Slack: see section 7.
- Other tools you connect through our tool layer (for example CRM or calendar tools): the data the agent needs to do the task you asked for. [LIST OF CONNECTORS AVAILABLE AT LAUNCH]
Data collected automatically
- Device and log data: IP address, browser type, pages viewed, timestamps, error reports.
- Usage and metering data: tokens and credits used, which model ran each task, agent actions and their receipts (what was proposed, approved, sent, and the result).
- Cookies and similar technologies: see our Cookie Notice (
cookie-notice.md).
Data from others
- Stripe: payment status, fraud signals, creator verification results.
- [ANY OTHER SOURCES, E.G. SIGN-IN PROVIDERS — CONFIRM]
We ask you not to upload special-category data (health, biometrics, religion, sexual life, criminal records and similar) unless you need to and the law allows it. See the Acceptable Use Policy (acceptable-use-policy.md).
<!-- REQUIRED | GDPR Art. 13(1)(c), 14(1)(d): purposes and categories of data | https://gdpr-info.eu/art-14-gdpr/ ; CCPA notice at collection, Cal. Civ. Code 1798.100 | https://cppa.ca.gov/regulations/ -->
5. How we use data and our legal bases
| Purpose | Data used | Legal basis (EU/UK) |
|---|---|---|
| Provide the service: run agents, keep your memory vault, carry out tasks you ask for, show receipts | Account data, customer content, connected-account data | Contract (Art. 6(1)(b)); for customer content we act on the customer's instructions as processor |
| Ask for approvals and carry out approved actions (for example sending an email) | Customer content, approval records | Contract |
| Billing, credits, invoices, tax | Account and billing data, metering data | Contract; legal obligation (Art. 6(1)(c)) for tax records |
| Marketplace: listings, purchases, creator payouts, ranking | Creator data, listing data, purchase records, buyer-confirmed results (never Gmail or Slack content) | Contract; legitimate interests in a fair, useful marketplace |
| Security, fraud and abuse prevention, enforcing our terms | Logs, account data, limited content review (see section 10) | Legitimate interests (Art. 6(1)(f)); legal obligation |
| Support | Support messages, account data | Contract; legitimate interests |
| Improve Cofello's own product (for example which features are used, error rates) | Usage and log data, **not** connected-account content | Legitimate interests; consent where cookies require it |
| Product and service emails | Email address | Contract (service messages); consent or legitimate interests for marketing, with an opt-out in every marketing email |
| Legal compliance and disputes | Any relevant data | Legal obligation; legitimate interests |
We do not make decisions about you based solely on automated processing that have legal or similarly significant effects. Marketplace ranking orders listings; it is not such a decision. If that changes we will tell you and offer human review.
<!-- REQUIRED | GDPR Art. 13(1)(c)-(d) legal bases and legitimate interests; Art. 22 automated decisions; Art. 13(2)(f) | https://gdpr-info.eu/art-22-gdpr/ -->
6. Google user data (including Gmail)
Limited Use statement. Cofello's use and transfer to any other app of information received from Google APIs will adhere to the Google API Services User Data Policy (https://developers.google.com/terms/api-services-user-data-policy), including the Limited Use requirements. The use of information received from Google Workspace scopes will adhere to the Google Workspace API User Data and Developer Policy (https://developers.google.com/workspace/workspace-api-user-data-developer-policy), including the Limited Use requirements.
<!-- REQUIRED | Google requires an affirmative Limited Use statement in the app or on a website page; the second sentence follows Google's own example wording | https://developers.google.com/workspace/workspace-api-user-data-developer-policy -->
<!-- REQUIRED | Google requires in-product disclosure before consent too, not only this policy. Code task: pre-consent screen. | same source -->
What we access. When you connect Gmail, you choose to grant Cofello these permissions: [LIST EXACT GOOGLE OAUTH SCOPES REQUESTED, E.G. READ, COMPOSE/DRAFT, SEND — NARROWEST POSSIBLE]. Depending on those permissions we can read email messages, headers, attachments and labels, create drafts, and, only with your approval or setting, send email from your account.
Why. Only to provide the user-facing features you see in Cofello: letting your agents read and summarise your email, find information you ask for, add relevant facts to your business memory vault, and draft or (with approval) send replies and outreach.
What we do not do with Google user data:
- We do not use it, or anything derived from it, to create, train or improve any machine-learning or AI model, other than a personalised feature used only by you in your own workspace (see the clone rule below).
- We do not transfer or sell it to advertisers, data brokers or information resellers, and we do not use it for ads.
- We do not use it to decide creditworthiness or for lending.
- We do not use it to benefit any other customer, to rank marketplace listings, or to give marketplace creators any information.
- Clones: a clone of you may draw on your own Gmail data only while it works inside your own workspace. Gmail-derived data is never included in a clone you list on the marketplace or license to others.
Who sees it. Your workspace's agents and the seats in your workspace (all seats share one memory vault; see section 11). Our staff do not read your email unless (a) you give us permission for specific messages, for example in a support request; (b) it is needed for security, such as investigating abuse or a bug; or (c) the law requires it. We log every staff access to message content.
Sharing. We pass Gmail data only to the service providers needed to run the feature you use (our hosting, database and tool-layer providers, and the AI model provider that processes a task), under contracts that bind them to these rules. See the Subprocessor List (subprocessor-list.md).
Storage and security. Google OAuth tokens are encrypted at rest [DESCRIBE KEY MANAGEMENT, E.G. CLOUD KMS — CONFIRM BEFORE PUBLISHING]. Data is encrypted in transit and at rest. See the Security Policy (security-policy.md).
Deleting it. You can disconnect Google at any time in [SETTINGS PATH] or at https://myaccount.google.com/permissions. When you disconnect, we stop accessing your account and delete stored Gmail message content and Gmail-derived memory entries within 30 days [RECOMMENDED — Chase to confirm], except copies in backups, which expire within 35 days [RECOMMENDED — Chase to confirm]. You can also ask us to delete it at [PRIVACY EMAIL].
<!-- REQUIRED | Limited Use rules: use only for prominent user-facing features; no transfers except listed; no human reading except consent/security/law; no ads, data brokers, credit; no AI/ML training beyond the specific user's personalised model | https://developers.google.com/terms/api-services-user-data-policy ; https://developers.google.com/workspace/workspace-api-user-data-developer-policy -->
<!-- LAWYER: the owner-clone line relies on Google's "beyond that specific user's personalized model" wording. Launch-blockers row 1 flags this as Medium confidence. Decide before clone launch. -->
7. Slack data
When you install the Cofello Slack app, we access the Slack data allowed by the permissions you approve: [LIST EXACT SLACK SCOPES]. We use it only to provide Cofello in your Slack workspace, for example to post approval requests, answer messages sent to Cofello, and give your agents context you asked them to use.
- We do not use Slack data to train any large language model or other AI model, under any circumstances.
- We do not use one Slack workspace's data to benefit any other organisation.
- We do not use Slack data to contact people outside Slack without their separate permission.
- We keep the minimum Slack data needed. We store only message references (such as channel and message IDs and links) and short summaries that a workspace owner approves into the memory vault. We do not keep copies of raw Slack messages beyond the time needed to complete a task, and never more than 24 hours. [RECOMMENDED — Chase to confirm]
- If you uninstall the Slack app, or we discontinue it, we delete the Slack data we hold within 14 business days.
<!-- REQUIRED | Slack Developer Policy: no LLM training; no contacting users outside Slack without separate permission; delete all data within 14 business days of uninstall; minimum scopes | https://docs.slack.dev/developer-policy/ -->
<!-- REQUIRED | Slack API Terms: no use of one org's data to directly benefit another; commercial distribution via Slack Marketplace | https://slack.com/terms-of-service/api -->
<!-- LAWYER/OWNER: Slack Marketplace AI guidelines say "don't store Slack data" for assistant apps; storage choice above is open (launch-blockers row 9). -->
8. AI model providers
Cofello is model-agnostic. Tasks are routed to third-party AI models chosen for quality, and you can pick a model yourself. To run a task, the relevant part of your request and context is sent to that model provider.
- We only send customer content to providers and service tiers whose terms say your inputs and outputs are not used to train their models, and that offer a data processing agreement. Providers we use are listed in the Subprocessor List.
- If you choose a model that is not approved for customer content, [CHOOSE: "we will not send connected-account or vault data to it" OR "we will warn you and ask before sending"].
- Some providers keep inputs for a short period for abuse monitoring (for example up to [PROVIDER RETENTION, E.G. 30 DAYS]) unless zero-data-retention applies. [CONFIRM PER PROVIDER]
- AI output can be wrong. Agents show their work and ask for approval before important external actions.
<!-- REQUIRED | Google Limited Use restricts transfers to what is needed for the user-facing feature; Slack bans LLM training | sources above -->
<!-- GOOD PRACTICE | routing allowlist per data class | launch-blockers row 12 -->
9. Clones
- Your own clone (owner clone): built only after you verify your identity and give explicit consent. We keep a record of that consent. You can pause or delete your clone at any time.
- Hired clones: a clone of another person that you hire from the marketplace works inside your workspace with your workspace's data. The clone's creator never receives your data, your memory vault, or your connected-account content.
- Clones of other people are only allowed with that person's verified, written consent. Clones of people who have died are not offered. A public page, recording or a family member's request is not consent.
- Clones are always labelled as AI (for example "AI clone of {Name} — not {Name}").
Clone material can include your voice, image and writing style. Where that counts as biometric data under a law that applies to you, we will ask for any additional consent that law requires. [LAWYER: CONFIRM WHETHER VOICE/FACE DATA IS PROCESSED AND WHETHER BIPA OR SIMILAR STATE LAWS APPLY]
<!-- GOOD PRACTICE + state law | consent and labelling; Tennessee ELVIS Act, Cal. AB 1836/AB 2602 noted in launch-blockers row 8 | https://leginfo.legislature.ca.gov/faces/billNavClient.xhtml?bill_id=202320240AB1836 -->
<!-- REQUIRED (if EU users) | EU AI Act Art. 50 — tell people they are interacting with AI | https://artificialintelligenceact.eu/article/50/ -->
10. Who we share data with
- Service providers (subprocessors) who host, store, route or process data for us, under contracts. Full list:
subprocessor-list.md, published at [SUBPROCESSOR PAGE URL].
- Services you connect (Google, Slack and others). When an agent acts in those services, they receive data under their own privacy policies.
- Stripe for payments and creator payouts (Stripe acts as an independent controller for some data, such as fraud prevention and identity checks: https://stripe.com/privacy).
- Marketplace creators receive only what they need to run their business on Cofello: sales counts, revenue, refunds and buyer-confirmed results. They never receive buyer content or connected-account data. [CONFIRM WHETHER CREATORS SEE BUYER NAMES/BUSINESS NAMES]
- Other seats in your workspace see the shared memory vault and agent activity.
- Legal and safety: when required by law or to protect rights, safety and security. We will tell the customer about a government request for customer content unless the law forbids it.
- Business transfers: in a merger, acquisition or sale of assets, data may transfer to the new owner, who must honour this policy. For Google user data we will get your explicit consent first.
We do not sell personal data and do not "share" it for cross-context behavioural advertising (as those words are defined in California law).
<!-- REQUIRED | GDPR Art. 13(1)(e) recipients; Google Limited Use merger exception requires explicit prior consent; CCPA "Do Not Sell or Share" disclosure | sources above -->
11. How long we keep data, and deletion (including the memory vault and seats)
| Data | How long |
|---|---|
| Account data | While your account is open, then 90 days after closure [RECOMMENDED — Chase to confirm], except billing and tax records and data under a legal hold |
| Customer content and memory vault | While the workspace is active. After the workspace is deleted or the contract ends: a 30-day export window, then deleted within 30 days [RECOMMENDED — Chase to confirm] |
| Gmail-derived data | Until you disconnect or delete it, then within 30 days [RECOMMENDED — Chase to confirm] |
| Slack data | Until uninstall, then within 14 business days |
| Billing, tax and payout records | 7 years after the end of the tax year; 10 years for EU VAT records if we register for the EU VAT One Stop Shop [RECOMMENDED — Chase to confirm] [ACCOUNTANT TO CONFIRM] |
| Security and access logs | 12 months [RECOMMENDED — Chase to confirm] |
| Agent action receipts in your workspace (who approved, when, recipients, subject, message ID; not the email body) | Life of the workspace; deleted with it [RECOMMENDED — Chase to confirm] |
| Cofello's send-compliance log (metadata only, for abuse prevention and anti-spam compliance) | 24 months [RECOMMENDED — Chase to confirm] |
| Records of privacy requests | 24 months [RECOMMENDED — Chase to confirm] |
| Backups | Rolling, expire within 35 days [RECOMMENDED — Chase to confirm] |
| Email suppression (unsubscribe) lists | Kept permanently, so we never email that address again |
| Clone consent and terms-acceptance records | For the life of the clone or account plus 3 years [RECOMMENDED — Chase to confirm] |
The shared memory vault. Each business has one memory vault. All agents, seats and clones in that business can use it. Separate businesses under one login are walled off from each other.
When a seat is removed:
1. The removed person loses access immediately.
2. Their connected accounts (for example their own Gmail) are disconnected and their tokens deleted.
3. Data that came from their own connected accounts (for example their Gmail, or their personal Slack user connection) is deleted from the vault within 30 days [RECOMMENDED — Chase to confirm], after a 7-day period in which the owner can undo a mistaken removal. The workspace owner cannot keep this data, because that person, not the owner, gave Google or Slack the consent for it. Data from a Slack app installed for the whole workspace is not affected by one seat leaving; it is deleted within 14 business days after the app is uninstalled.
4. Business knowledge that the owner or other seats added themselves, and content the person created in the workspace for the business (chats, notes, documents), stays with the business, because the business is the controller. The person can ask the business to delete or correct it.
Deleting an individual item. Owners can correct, archive or permanently delete memory entries. [CONFIRM PERMANENT DELETE IS SHIPPED — FOUNDER DOC SAYS IT WAS DEFERRED]
Deleting your account or workspace. In [SETTINGS PATH] or by email to [PRIVACY EMAIL]. We delete or anonymise data across our systems and ask our subprocessors to do the same, within the periods above. Backups are not edited item by item; deleted data disappears from backups when they expire (35 days) and is never restored into live systems in the meantime. We may keep data longer only where the law requires it or for an active legal claim, and only for that purpose.
<!-- REQUIRED | GDPR Art. 12(3): act on requests within one month (extendable by two months) | https://gdpr-info.eu/art-12-gdpr/ ; ICO: erasure must cover backups and individuals must be told what happens | https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/individual-rights/individual-rights/right-to-erasure/ ; CCPA regs §7021(b) 45 days, §7022(d) backups, §7101 24-month records | https://www.law.cornell.edu/regulations/california/11-CCR-7021 ; https://www.law.cornell.edu/regulations/california/11-CCR-7022 ; https://www.law.cornell.edu/regulations/california/11-CCR-7101 ; IRS record periods | https://www.irs.gov/businesses/small-businesses-self-employed/how-long-should-i-keep-records ; EU VAT Directive Art. 369 (10 years) | https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:02006L0112-20250101 -->
<!-- REQUIRED | GDPR Art. 5(1)(e) storage limitation, Art. 13(2)(a) retention periods, Art. 17 erasure | https://gdpr-info.eu/art-17-gdpr/ ; Google requires helping users delete data | https://developers.google.com/workspace/workspace-api-user-data-developer-policy -->
12. Security
We use technical and organisational measures to protect data: encryption in transit and at rest, encrypted credential storage, access limited to staff who need it, logging of staff access to customer content, and workspace isolation. No system is perfectly secure. Details: security-policy.md. Cofello does not currently hold SOC 2 or ISO 27001 certification; [IF APPLICABLE: "an annual Google CASA assessment is required for Gmail access and is STATUS"].
If a personal data breach affects you, we will tell you and the relevant authorities as the law requires (for example, EU and UK authorities within 72 hours of becoming aware, where required).
<!-- REQUIRED | GDPR Art. 32-34 | https://gdpr-info.eu/art-33-gdpr/ -->
13. International transfers
We are based in the United States and our providers process data in [LIST HOSTING REGIONS, E.G. US]. When we move personal data from the EU, EEA, UK or Switzerland to countries without an adequacy decision, we use the European Commission's Standard Contractual Clauses (https://eur-lex.europa.eu/eli/dec_impl/2021/914/oj), the UK International Data Transfer Addendum, and/or the EU-US Data Privacy Framework where the recipient is certified. [STATE WHETHER COFELLO SELF-CERTIFIES TO THE DPF — CURRENTLY ASSUMED: NOT CERTIFIED] You can ask us for a copy of the safeguards.
<!-- REQUIRED | GDPR Art. 13(1)(f), Art. 44-46 | https://commission.europa.eu/law/law-topic/data-protection/international-dimension-data-protection/standard-contractual-clauses-scc_en ; UK Addendum | https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/international-transfers/appropriate-safeguards/what-are-standard-data-protection-clauses-the-uk-idta-and-the-addendum/ -->
14. Your rights
Everyone. You can access, correct, export or delete your data, and disconnect connected accounts, through [SETTINGS PATH] or by writing to [PRIVACY EMAIL]. We will reply within one month (EU/UK) or 45 days (California and similar US state laws), and we will tell you if we need more time where the law allows an extension [RECOMMENDED — Chase to confirm].
EU, EEA and UK. You have the right to access, rectify, erase, restrict, port, and object to processing based on legitimate interests or for direct marketing; to withdraw consent at any time; and to complain to a supervisory authority, for example where you live or work. Our representative in the EU is [EU ARTICLE 27 REPRESENTATIVE NAME AND ADDRESS] and in the UK is [UK REPRESENTATIVE NAME AND ADDRESS].
California and other US states. Depending on your state you may have the right to know, access, correct, delete, and opt out of sale, sharing, targeted advertising and certain profiling. We do not sell or share personal data and do not use it for targeted advertising. We will not discriminate against you for using your rights. You may use an authorised agent. We verify requests by [VERIFICATION METHOD]. [LAWYER: CONFIRM WHICH STATE LAWS APPLY GIVEN COFELLO'S SIZE; CCPA THRESHOLDS MAY NOT YET BE MET]
People whose data is in a customer's workspace (for example email correspondents): contact that customer. If you contact us, we will forward your request to the customer.
<!-- REQUIRED | GDPR Arts. 12-22, 27, 77 | https://gdpr-info.eu/art-27-gdpr/ ; CCPA | https://cppa.ca.gov/regulations/ -->
15. Children
Cofello is for businesses and people 18 or older. It is not for children. We do not knowingly collect data from anyone under 18 (or under 16 in the EU/UK, where the age of digital consent may be lower than 18). If we learn we have, we delete it. Tell us at [PRIVACY EMAIL].
<!-- REQUIRED | COPPA (actual knowledge of under-13s) and GDPR Art. 8 | https://www.federalregister.gov/documents/2025/04/22/2025-05904/childrens-online-privacy-protection-rule ; https://gdpr-info.eu/art-8-gdpr/ ; Google bars child-directed apps from Google sign-in/data | https://developers.google.com/terms/api-services-user-data-policy -->
16. Marketing emails
We send product and account emails you need. We send marketing emails only where the law allows, and every one has an unsubscribe link. Cofello never uses your connected Gmail or Slack data for our own marketing.
17. Changes to this policy
We will post changes here and update the date above. If a change is material, or if we want to use Google user data in a new way, we will tell you in advance and ask for your consent where required, before the change applies.
<!-- REQUIRED | Google: notify users and get consent before using Google user data in a new way | https://developers.google.com/terms/api-services-user-data-policy -->
18. Contact
[COMPANY LEGAL NAME], [REGISTERED BUSINESS ADDRESS]. Privacy: [PRIVACY EMAIL]. Security: [SECURITY EMAIL]. EU representative: [EU ARTICLE 27 REPRESENTATIVE]. UK representative: [UK REPRESENTATIVE].
DRAFT — not legal advice — lawyer must review before launch