Queryable

AI tool infrastructure / in development

Give agent tools an operating layer.

Queryable is developing infrastructure for the point where an AI tool meets a real system: connections, execution, authority and deployment.

Explore the architectureDiscuss a use case

Interactive architecture illustration

order.statuspermission boundary
callersupport_assistantscopeorders:readoperationget_order_statusresourceorders / tenant_42
ALLOW

The requested read is within this caller’s example scope.

Illustrative permission logic running in your browser. No live system is called.

From intent to an accountable operation

  1. 01 / IntentAgent or application
  2. 02 / BoundaryIdentity + permission
  3. 03 / ExecutionA scoped tool
  4. 04 / ResourceA real system

The work around a tool call

Define once.
Operate deliberately.

01 /

Connections

Which resource can this tool reach?

02 /

Authority

Who may perform this operation?

03 /

Execution

What happens when it fails?

04 /

Reuse

Who owns and maintains the boundary?

A concrete starting point

An order update. A defined scope.

A support assistant needs one answer. Define the accepted identifier, permitted fields and tenant scope before exposing the operation.

Read the developer notes
Example tool brief / conceptual
operation: get_order_status
resource: orders
permission: orders:read
tenant: from_authenticated_context
input: order_id
output:
  - status
  - estimated_delivery
exclude:
  - payment_credentials
  - unrelated_customer_records

Where to begin

One operation. Clear boundaries.

Bring one concrete operation

What should your tools be allowed to do?

Discuss your use case