How your data is kept
Workspace access controls, private file storage and encrypted Gmail credentials protect different parts of the service. This page describes those controls and their limits; it is not a security certification.
Last updated 20 October 2026.
Tenant isolation
Isolation is a database property here, not an application convention. Each operational table carries an organization id and a row-level security policy; the application connects as a least-privilege role whose grants are pinned to a reviewed manifest, and each transaction sets the tenant claim inside the transaction itself.
Before release, our checks must confirm that every tenant table has an isolation policy and the reviewed runtime-role grants. Release review also requires evidence for each instance, including fail-closed behaviour for unbound requests on the US instance and runtime checks that one organisation cannot read another's data. These are release requirements; this page does not attest that hosted continuous-integration checks have executed for a given release.
What the AI is allowed to do
- Propose, never write
- Extraction, matching and sourcing produce proposals. A person approves, corrects or rejects before anything becomes a record or leaves the building.
- Grounded fields only
- Every extracted field carries an evidence span from the source document. Ungrounded fields are dropped and counted.
- Rules before models
- Catalog matching runs deterministic rules first; the model only re-ranks what rules could not place, and abstains below the tenant's threshold.
- Tool tiers
- Untrusted text is handled by tool-less calls. Tools are available only to calls whose input is trusted, enforced in the type system.
- Budgets
- Per-tenant spend has hard stops. A runaway job halts instead of billing.
The ledger
The model-call ledger records prompt hash, model, tokens, cost, confidence, injection flags, tenant and purpose. You can see what the AI did on your data, what it cost, and how sure it was — per call, not as a monthly total. Two kinds of call are not in the ledger yet: the supplier web search, and the TypeSafe judgments used on the reference customer's instance.
Secrets and access
Provider client secrets are held in environment configuration. Connected Gmail refresh credentials are encrypted in the database and bound to the mailbox and issuing client. Disconnect removes the saved Gmail credentials; it does not erase imported business history. Google-derived content is subject to the access and Limited Use commitments on our privacy page.
Private originals are served through authorized application requests. A file being stored privately does not mean every attachment has been malware scanned. Deleting a generated preview is distinct from deleting its original and message record.
What is not true yet
We are pre-launch and say what is missing: no SOC 2 report, no penetration-test letter to share, each instance hosted in a single region, and a support model that is a person rather than a rota. On the list above: fail-closed tenancy and the two-organisation isolation harness are staged but not yet the default; per-tenant envelope encryption of secrets is the next migration; the evaluation set exists but does not yet gate releases; and the support audit trail is not yet visible to tenants. If any of those is a blocker for your pilot, tell us at the first call rather than after.
A working Google connection and this policy page do not establish Google scope verification or a completed CASA security assessment. Questions that need a real answer, not a page: write to hello@bizily.ai.
Questions
- Are you SOC 2 certified?
- No. The system is designed against SOC 2 criteria and the controls are mapped to them, but there has been no audit and we do not claim a report. We will say so plainly when that changes.
- Can AI features process my correspondence?
- Yes. Authorized workspace AI features can use shared conversation history, including imported Gmail correspondence. Connecting Gmail does not automatically submit every Sent message to a model. Gateway calls governed by a workspace's zero-data-retention policy request zero retention and prohibit prompt training. Other paths remain subject to their disclosed handling and workspace restrictions. We do not use Google Workspace API data to train generalized or non-personalized models, or authorize providers to do so. See Privacy for the Limited Use commitments.
- Can another tenant see my prices or customers?
- No. Every operational table carries an organization id with a row-level security policy, and the runtime database role is least-privilege. Fail-closed behaviour when a request carries no tenant claim is staged and becomes the default before any outside tenant is onboarded.
- What happens if a request contains a prompt injection?
- Untrusted text is only ever passed to tool-less, structured-output calls, wrapped in a per-call random delimiter with a canary. A canary echo or delimiter breakout flags the document, forces human review and shows a banner. Injection cases are part of the release benchmark.
- Who at Bizily can see my tenant?
- Workspace colleagues access shared records according to their permissions. Personnel and service-provider access to Google-derived content is limited to documented user permission for the specific information, security needs, legal obligations or another circumstance expressly permitted by Google's Limited Use requirements. A tenant-visible support read audit trail is being added.
- Where does the data live?
- The current app.bizily.ai workspace uses the existing ManaEsse backend and integrations; the legacy ManaEsse hostname reaches that same workspace. Its application and database are hosted in Singapore. Files, email and enabled external services have their own processing locations. A separate deployment may use a different region and integrations; agree the applicable setup before a pilot. See /subprocessors for the current services.
- Can I export or delete everything?
- Workspace export is available. Disconnect removes saved Gmail credentials; Archive hides a conversation and can be undone. Neither operation erases imported history or originals, and closure is not confirmation that all copies were deleted. Automated organization-wide permanent purge is not currently enabled. Contact hello@bizily.ai for a request; we verify its scope, explain retention or technical limits and confirm the action actually completed.