Give your AI agent autonomy,not your keys.

No more cobbling together your own safeguards: your agent deploys your apps, and it doesn’t delete your production database without your approval.

Talk to the team

A conversation with your agent

You: Deploy the app, and connect it to our payment tool.
Your agent: It’s live. Enter the payment key in while: I will never see it.
You: Delete the old production database.
Your agent: Requested. The deletion is waiting for your approval. Delete the production databaserequested by your agent to approve

Works with

Click your agent to copy its instructions.

Hosted in Franceby Nextmap

Everything is in place from the first deployment

Keep your keys out of your agent’s reach

You enter an external API’s key in while. Your app uses it without ever reading it, and your agent never sees it.

A write to an external API is refused until a person has approved its rule.
After 7 days, only the calls seen during that period go through.
A leaked key is replaced without changing a line of code.

Decide what needs your approval

By default, your agent deploys without waiting for you. The most sensitive actions always wait for a person, and everything is recorded in a signed activity log.

Deleting or restoring a production database.
Raising a budget or changing a rule.
An API call you marked “approve every call”, like a refund.

Isolate each app and fix its vulnerabilities

Each app runs in its own sandbox, and a known serious vulnerability warns your agent, which can update the dependency.

Over the internal network, an app can’t reach any app in another project.
Its outbound traffic to the internet is closed by default.
Each version’s dependencies are listed (SBOM) and checked against known vulnerabilities at every deployment, then every day.

Open each app to whoever you choose

An app opens to everyone, with a password, or through your company account.

On the Pro and Enterprise plans, the company account goes through OpenID Connect: Microsoft Entra, Google Workspace, Okta, Keycloak or another compatible provider.
Everything goes over HTTPS, with HSTS by default.

Roll back, in the code and in the data

Your agent puts the previous version back online, and your database goes back to a precise moment.

Each database is backed up every day, and in production, one backup a day is tested by a real restore.
Backups are encrypted with AES-256 and kept for 14 days.

Keep your data in France, with a French company

Your apps and their backups are hosted in France, by Nextmap.

Fill in your security questionnaire

Builds: Each build runs in its own small virtual machine.
Secrets and configuration: Secrets left in the code are detected, and public addresses are checked from the outside after each production deployment, then every night.
Activity log: It is append-only, chained, and signed by a key that never leaves its vault. It can be exported and re-verified with a small script. With the Enterprise plan, it also streams to your SIEM.
Backups: They go to storage outside the cluster where the apps run, dedicated to the organization and versioned.
Cloud access: A production app reaches Google Cloud or Azure without any key, through OIDC identity federation.
Cutting access: You disable a service account, revoke a key or a token, close a person’s sessions or remove an access, at any time.
In case of intrusion: The while team can suspend builds and deployments in one action, and revoke access, without shutting down the apps that are running.

Questions

Who can raise my agent’s budget?

A person raises it, never an agent.

Can one app read another app’s data?

Only the database of another app in your organization, in production, read only, and with the approval of a person who manages that database.

What if my agent is compromised?

You disconnect it from while at any time, and its tokens stop working.

How do I report a vulnerability in while?

Write to security@while.sh. The address is also in the site’s security.txt file.

A security question? Ask the team

The team’s address

hello@while.sh