Detectable changes
Security events and evidence summaries chained by hash, with anchors published outside the platform.
Security and governance
Protect your data and prove what happened.
Define who accesses what, keep AI within your company's permissions and give auditors evidence they can check outside the platform.
Five layers protect your operation.
Legal hold
Keep data preserved for legal purposes, above retention rules.
Your company's login
Sign in with the company's identity provider through SAML or OIDC, create and deactivate people through SCIM and sync the directory through LDAP.
Role-based access
Define what each profile does, by module and by action. What a person can't see is neither shown nor counted.
Isolated data
Each company's data is isolated by row-level security, in the database itself.
Give the auditor evidence they can check outside the platform.
-
Record 1
own hash
-
Record 2
+ previous hash
-
Record n
+ previous hash
-
Anchor
locked retention
Chained
Each record carries the hash of the previous one. Changing one breaks the chain from there on.
Anchored
On SaaS, anchors are written to storage with locked retention. In a self-hosted installation, the destination is defined with the organization.
Verifiable
Exported evidence can be recalculated by independent auditors, outside the platform. The link to the anchor is checked in the trail.
Keep AI within your company's permissions.
Permission is resolved in the database query, before anything is sent to the model. In conversation, the agent receives only the data the requester can access. Each automatic plan follows its owner's permission.
Identity
When a person asks, the agent acts with that person's role and permission. Automatic work stays recorded and identified in the trail.
Permission on the data
The same check the screen does applies to the agent, enforced at the data layer.
Trail
Each agent action enters the audit trail, with who asked and what was done. Documents consulted are recorded with their version.
Confirmation
In conversation, every action that writes goes through a confirmation card, with values frozen until accepted. Text received by email or in a document doesn't authorize the agent to act.
Choose where the platform runs.
SaaS
Managed cloud
Infrastructure on Google Cloud, currently in the São Paulo region, with per-company isolation, daily backup and the platform operated by F7 KORE.
On-premises
On your company's infrastructure
Platform, database, files and trail on the organization's machines, through a deployment project. Installation requires outbound network access to the image registry and the language model provider.
AI models
The platform uses Anthropic and Google models. On SaaS, configuration is handled by F7 KORE; on a dedicated instance, calls can use the organization's account with those providers. Only the content needed for the task and authorized goes to the model. Calculations and forecasts run in the database.
Answer audits and regulators with evidence.
Data protection
Traceability of automated decisions, to support the human review provided for in data protection laws.
Use of artificial intelligence
A record of who asked, what was done and which documents the agent consulted.
Audit
Evidence for internal and third-party audits, verifiable outside the platform.
Traceable decisions and actions, from the source of each number to the person who owns the fix.
Talk to our team and see what F7 KORE can do for your operation.
Product screens and values use demonstration data.