Governance and security
Every agent under your rules. Every action on record.
The controls behind every agent, how we protect your data, and the documents your review needs.
Hosted in the EU. A data processing agreement on every plan. No training on your data.
An agent wants to actSend an email, pay an invoice
- Allowed?A rule must allow the app and action
- Approved?An employee says yes, if you require it
- Within budget?Paused at the limit
- On recordLogged with its decision
Controls
What a security review looks at.
Five areas, and how Blentera handles each.
| Owner | Admin | Operator | Viewer | |
|---|---|---|---|---|
| View work | Yes | Yes | Yes | Yes |
| Approve actions | Yes | Yes | Yes | – |
| Assign tasks | Yes | Yes | Yes | – |
| Set up agents | Yes | Yes | – | – |
| Connect apps | Yes | Yes | – | – |
| Invite employees | Yes | Yes | – | – |
| Manage roles | Yes | – | – | – |
Only the access the job needs.
Four roles, from owner to viewer, with passkeys and two-step verification on every plan and single sign-on on Business. Agents get nothing by default: you choose which apps each one may use, and set every action to on, ask first or off.
An employee says yes first.
Set rules per agent, project, app or tool, such as every payment or customer email. Only employees approve, never an agent, and each approval covers the exact action reviewed. Simple approvals also work in Slack and Microsoft Teams.
Owner warned at 80%
Paused at the limit
A ceiling on spending. A stop at any time.
Budgets per company, agent and project warn at 80% and pause the work at the limit, until an employee decides. Pause any agent, or stop a run and its right to make changes at once.
- 09:12Support agentRead a customer email · Allowed
- 09:14Finance agentPaid an invoice · Approved by an employee
- 09:20Sales agentSend an offer · Waiting for approval
- 09:31Marketing agentDelete a contact · Blocked by a rule
Every action, on record.
Every action by an agent or an employee, with its source, its approval and the rule behind it. Keys and passwords never appear in the log. Filter it, and export the evidence for an auditor.
- Platform and databaseEU
- Files and backupsEU, encrypted
- Keys to your appsAES-256-GCM
- Every connectionTLS
- Database addressPrivate
Your data stays in the EU, and stays yours.
Hosted in the EU, with encrypted connections, keys and backups, and exportable at any time. Agents use the AI provider you connect, under your agreement, and Blentera never trains AI models on your data.
EU AI Act and GDPR
Checked before it goes live.
Your company stays accountable for its use of AI. Blentera runs the checks, blocks what fails and keeps the evidence.
- SupportCustomer repliesLive
- FinanceSupplier invoice checksLive
- HRScreening job applicantsBlocked
- SalesWeekly sales summaryReview due
- LegalContract first draftsAssessment pending
Three assessments
Each use case is assessed against the EU AI Act, the GDPR and your own policy before it goes live.
Uses we block
Use cases that decide about individuals, such as screening job applicants, cannot go live. Nor can one your assessment marks as high-risk or prohibited.
Reviews that expire
Every use case has an owner and a review date. Once the review is overdue, it cannot go live.
Shared responsibility
Who does what.
Blentera enforces your rules. A few decisions stay with your company.
Roles, rules and budgets
Blentera: Enforces them on every action
Your company: Sets them, and chooses who approves
Actions that are hard to undo
Blentera: Holds them until an employee says yes
Your company: Decides which actions need a yes
AI providers and regions
Blentera: Connects only the ones you choose
Your company: Chooses what fits its own rules
AI use cases
Blentera: Runs the checks and blocks what fails
Your company: Makes the final assessment
Records and transparency
Blentera: Keeps the activity log and the evidence
Your company: Tells employees and customers where AI is used
Documents
Everything your review needs.
Read them here, or send us your security questionnaire and we answer it directly.
- Data processing agreementGDPR Article 28 terms, on every plan.ReadGet a signed copy
- Sub-processorsWho processes data for us, what and where.Read
- Privacy policyPersonal data we process as controller.Read
- Terms of serviceThe agreement for using Blentera.Read
- Service statusLive status of the platform.Open
- Security contactOur security.txt, for researchers and tools.Open
- Security questionnaireSend us yours. We answer it directly.Email us
Passkeys for our operators
Blentera's platform operators can sign in only with a passkey.
Agent servers apart
Each Blentera agent server serves one company and has no public address.
Breach notice within 48 hours
We tell you within 48 hours of becoming aware of a breach that affects your data.
For your review
Questions security teams ask.
Short answers to what reviews ask most. Anything missing? Write to us.
- In the EU, hosted by UpCloud, a European cloud provider. Enterprise customers can run Blentera on their own servers.
- The platform and your data stay in the EU. If a sub-processor processes personal data outside the European Economic Area, an adequacy decision or the European Commission's standard contractual clauses cover it. AI providers you connect may process data elsewhere, under your agreement with them.
- Access is limited to what is needed to run and support the service. If a support case needs your company's data, we ask a company owner first.
- Yes, when they use your apps through Blentera. Approvals, budgets and the log then apply to OpenClaw, Hermes and Cursor Cloud agents. What an outside agent does with its own tools, outside Blentera, is not covered.
- UpCloud for hosting, database, files and agent servers, and Mailgun for email, both in the EU. The full list is in our data processing agreement. We give notice before any change, so you can object.
- Export it at any time. When your subscription ends or you delete your company, we delete the personal data we process for you, company memory included. Copies in backups are removed when those backups reach their deletion date.
- Text in an email or a document can try to steer an agent. Access rules, approvals, budgets and the activity log limit what any agent can do, whatever it is told.
- Live status is on our status page. An uptime agreement is part of the Enterprise plan.
- There is no automatic time limit. Records stay for as long as your company uses Blentera and are deleted with the company, or with the agent they belong to.
- Not yet. Every release is checked for known vulnerabilities in its dependencies and runs our security regression tests. This page and our data processing agreement describe the measures in place, and we answer security questionnaires directly.
- No tool can do that alone. Your company is the deployer and makes the final assessment. Blentera gives you the controls the rules ask for, such as human oversight, logging and a use case register, and keeps the evidence.
- Write to hello@blentera.com with the details and steps to reproduce. Do not test against other customers’ data, and give us reasonable time to fix an issue before you share it publicly.
Get started
Put your first agent to work in minutes.
Three steps, and your first agent is working for your company.
Step 1
Create an account
Sign up with your work email and invite your team. No sales call needed.
Step 2
Add your first agent
Create an agent, give it a role, and connect your own AI key and the apps it may use.
Step 3
Give it a task
Ask in Blentera or your chat app, and follow the work until it is done.