Honeycomb Privacy Policy

Effective date: [Effective date] Last updated: [Last updated]

1. Summary

• Scaled Abundance Inc. ("we", "us") runs Honeycomb, a workspace where a company keeps its Company Brain (what the business knows), connects its business systems, and runs AI agents that propose work for people to approve.
• When a company uses Honeycomb, the company decides what business data goes in. For that business data we act as the company's processor (service provider). For the account data of the people who sign in, and for running our own business, we act as controller.
• Connected systems are read live through connectors. We keep a record of each connector call, but not the data values in it. Connection credentials are stored encrypted.
• Text you give Honeycomb is sent to AI model providers to produce answers and drafts. The main provider is Anthropic. Some features use OpenAI. A company may choose another provider.
• Neither we nor our model providers use your data to train models. Provider retention follows each provider's standard API data retention terms.
• Actions that change data in your connected systems need a person's approval unless your company has set a rule that approves that kind of action automatically.
• We do not sell personal data. [Confirm: we do not sell or share personal data for cross-context behavioral advertising.]
• We do not use advertising or analytics cookies. See our Cookie Policy.

2. Who we are

Honeycomb is provided by Scaled Abundance Inc., a Delaware corporation operating out of Texas, [Registered address], registration number [Company registration number]. You can reach us at info@scaledabundance.ai. Our data protection contact is info@scaledabundance.ai. [If required: our representative in the EU / UK is [EU/UK representative].]

3. Our two roles

Controller. We decide how account data is used (your name, email, sign-in records, billing details, support messages) and how we run, secure and improve the service.

Processor. Business data that a customer company puts into Honeycomb or connects to it (its org chart, Company Brain knowledge, uploaded files, records read from connected systems, mailbox content it chooses to share, chats with agents) is processed on that company's instructions. The customer company is the controller of that data. If you are an employee or contact of a customer company and have a question about that data, please contact that company first. Our processing terms are in the Terms of Service, section 9, and in [Data Processing Addendum URL].

4. Data we hold

4.1 Account data (we are controller)

• When you sign up with email and password: your name, email address and a password, which is stored as a hash by our authentication library, never in plain text.
• Email verification: whether and when you verified your email address.
• Invite codes: sign-up is by invitation on our hosted service. We record which invite was used to create your account.
• Single sign-on: if your company signs you in through its own identity provider (OpenID Connect), we receive your name and email from that provider and confirm your email is at the company's domain. We do not offer sign-in through social networks.
• Sessions: for each signed-in session we store a session token, an expiry time, your IP address and your browser's user agent string, and which company you are working in.
• Team membership: the companies you belong to, your role, invitations you sent or received, and records of when a member left or was removed.
• Two-factor data: where two-factor sign-in is used, an encrypted authenticator secret and backup codes. [Confirm whether this applies to the customer-facing app.]

4.2 Company data (we are processor)

• Organization and org chart: company name and settings, departments, positions, roles, goals and key results.
• Company Brain knowledge: facts, rules, decisions and relationships the company records or that Honeycomb learns from interviews, agent runs, files and connected systems, plus their change history.
• Uploaded files: files you add to the Company Brain (for example PDF, Word, Excel, CSV, text, Markdown, PNG and JPG). File contents are encrypted with a key specific to your company.
• Conversations: chats with Honeycomb and its agents, setup interviews, and agent run steps.
• Approvals: actions an agent proposed, who approved or denied them, any note, and what happened when an approved action was sent.
• Feedback you send us: your note, the page and link you were on, the screen element you pointed at, your screen size and the app version.

4.3 Connector data (we are processor)

• Credentials: API keys and OAuth tokens for the systems your company connects. These are encrypted before they are stored.
• Live reads: when an agent, a person or an approved app needs data from a connected system, Honeycomb reads it from that system at that moment.
• Call records: for each connector call we store facts about the call only: which connection and tool, read or write, status, error type, timing, size, record count, and who or what made it. We do not store the request or response body, any field values, or any credential in these records. Call records are kept for 90 days by default; owners and admins can change this from 1 to 3,650 days.
• Mirrored reference records: for record types a connector declares, Honeycomb keeps a reference copy in the Company Brain (an ID, a title and simple attributes, plus whether the record is still active) so knowledge can link to the right customer, order or account. [Confirm which record types and attributes each catalog connector mirrors.]
• Mailboxes, files and calendars (Google Workspace, Microsoft 365): if your company turns these on, a person can connect their own mailbox, or an admin can connect a shared one. Tokens are encrypted. Message details become company knowledge only if the connection is set to share with the brain (off by default), and only for threads someone points the brain at. Agents may use a mailbox only if allowed (off by default), and by default they only draft.
• Slack: if connected, the bot token and signing secret are encrypted. By default, personal and financial identifiers are masked in what Honeycomb posts to Slack.
• Demo accounts: we may let a new company connect to shared demonstration accounts (for example sample CRM, inventory or accounting systems) that contain only fictional data. A demo connection holds no credentials of its own.

4.4 Usage and spend records

For each AI model call we record the company, the person whose action caused it (when there is one), the feature and purpose, the model, the number of tokens in and out, and the cost. We use these to show your company its spend, to enforce daily and monthly spending limits, and to bill.

4.5 Billing data

If you buy a paid plan, payment is handled by Stripe. We store your Stripe customer and subscription IDs, your plan, its status and the current period. We do not receive or store your full card number.

4.6 Technical data and logs

Our hosting provider keeps server logs. Error reports we write contain the error type, a message cut to 500 characters with personal details scrubbed, and short context values; they do not include request bodies or secrets. We also record your browser's time zone for your company when its time zone has not been set. [Confirm log retention period at our hosting provider: [Log retention period].]

5. How AI processing works

• To answer questions, run agents, read files and build knowledge, Honeycomb sends the relevant text (your prompt, relevant company knowledge, and data read from connected systems) to an AI model provider.
• By default this is Anthropic (Claude models). Honeycomb also supports OpenAI, Google (Gemini), Mistral, DeepSeek and Groq, which a company can choose in its settings, optionally with its own API key (stored encrypted). [Model providers to confirm: which providers are enabled on our hosted service.]
• Search across the Company Brain uses embeddings made by OpenAI (text-embedding-3-small), whichever provider the company chose for answers.
• No training on your data. Our model providers do not use the data we send them to train their models, and we extend the same commitment to you: we do not use customer data to train models.
• Provider retention. Each model provider keeps API inputs and outputs only under its standard API data retention terms: see [Anthropic API data retention terms] and [OpenAI API data retention terms]. [If a company chooses another provider, that provider's standard API data retention terms apply: link to confirm.]
• AI output can be wrong. That is why actions that change your systems go through human approval (see the Terms of Service, section 8).

6. Why we use data (and legal bases)

[Confirm: whether we send marketing email, and if so the consent basis and unsubscribe method.]

7. Who we share data with

• Hosting and database: Vercel (application hosting, scheduled jobs and logs) and Neon (Postgres database). [Hosting regions to confirm.]
• AI model providers: as listed in section 5.
• Payments: Stripe.
• Fonts: our fonts are served from our own servers; loading a page does not send your IP address to a font provider.
• Email delivery: [Email provider to confirm]. Our hosted service does not yet send email through an outside provider.
• Systems your company connects: Honeycomb sends requests to them on your company's behalf.
• Apps built by the Software Factory: if your company approves such an app, it can reach only the systems and permissions your company approved (see Terms, section 10).
• Authorities and advisers: when the law requires it, or to protect rights and safety; and professional advisers under confidentiality.
• Business transfers: a buyer or successor, under this policy.

The current list of subprocessors is at [Subprocessor list URL].

8. How long we keep data

• Connector call records: 90 days by default, adjustable by the company from 1 to 3,650 days, then deleted automatically.
• Sessions: a session expires 7 days after it starts.
• Single-use links: password reset links expire after 1 hour; connection sign-in state expires after 10 minutes.
• Company Brain knowledge: kept while the company account is active. Knowledge is protected against accidental deletion; owners can permanently delete selected knowledge, and deleting the company removes all of its knowledge in one step, leaving only a deletion record with counts and fingerprints, not content.
• Uploaded files: when deleted, the file contents are removed; a record of the file name and what was learned from it stays unless the owner also chooses to forget what was learned.
• Account data: [Retention period for account data after account closure].
• Company data after the contract ends: [Data retention after termination, e.g. deleted within N days after an export window].
• Billing records: [Billing record retention, e.g. as tax law requires].
• Logs: [Log retention period].

9. Security

• Credentials, OAuth tokens, customer-supplied model API keys and Slack secrets are encrypted with AES-256-GCM before storage.
• Uploaded files are encrypted with a per-company key, and a file copied to another company cannot be decrypted.
• Session cookies are HTTP-only, sent only over HTTPS in production, and limited to same-site use.
• Sign-in attempts and password resets are rate limited.
• Apps built by the Software Factory get short-lived tokens (15 minutes), and we store only a hash of each key and token.
• Invite codes are stored as hashes.

No system is perfectly secure. Please use a strong password and tell us at info@scaledabundance.ai if you suspect a problem.

10. Your rights

Depending on where you live, you may have the right to access, correct, delete, export or restrict the use of your personal data, to object to processing, and to withdraw consent. Company owners and admins can export the Company Brain as files. To make a request about your account data, write to info@scaledabundance.ai. For data a customer company controls, we will pass your request to that company and help it respond. You may also complain to your data protection authority. [Confirm: self-service account deletion; today account removal is handled on request.]

[California and other US state notices: [US state privacy notice to confirm].]

11. International transfers

Our providers may process data outside your country, including in the United States. Where required we rely on safeguards such as Standard Contractual Clauses. [Transfer mechanism to confirm.]

12. Children

Honeycomb is a business service and is not meant for anyone under [Minimum age]. If we learn we hold a child's data, we will delete it.

13. Changes

We will update the date at the top when this policy changes and tell account owners about material changes by [Notice method] at least [Notice period] before they take effect.

14. Contact

Scaled Abundance Inc., [Registered address]. Privacy questions: info@scaledabundance.ai. Data protection contact: info@scaledabundance.ai.