callbound.
Open workspace
CONTROL PLANE / AGENT EXECUTION

Give every agent
a perimeter.

Define what an onchain AI agent can call, see what each path may cost, and follow every decision back to its rule.

Build a policy Explore the protocol
QUICK POLICY CHECKRead market data / ready
01 / GATECapabilityAllowed methods only
02 / METERBudgetBefore execution
03 / PROOFTraceEvery decision logged
REQUEST VECTOR → POLICY RESULT
01 / THE WORKSPACEINTERACTIVE LOCAL DEMO

Set the rules.
See the consequences.

Configure a sample policy, compare modeled cost, and run requests through the same visible checks. Demo figures are illustrative and do not use a live network.

POLICY WORKSPACELOCAL SIMULATION
POLICY / 001
REQUEST BUILDER

Test the boundary

Ready to simulate

Choose an action and estimated workload. The policy checks capability first, then available daily budget.

ROUTEBalanced route
MODELED COST$0.38
AVAILABLE TODAY$12.00
Change the policy or run a request to inspect its decision.
DEMO ENGINE / NO WALLET OR LIVE EXECUTIONCAPABILITY → BUDGET → DECISION → TRACE
02 / THE CONTROL LOOPDESIGNED FOR EXPLICIT EXECUTION

Trust the route.
Inspect the rule.

Autonomy becomes useful when its limits are legible. Callbound gives each request a small, accountable path.

01

Grant the minimum

Define allowed actions per agent. A read request and a transaction request never inherit the same permission.

↗
02

Price before running

Compare modeled execution paths against a daily ceiling before an agent spends a unit of budget.

↗
03

Explain the outcome

Keep a decision trail with the capability check, budget check, and route that produced it.

↗
03 / BUILDER VIEW

A boundary you
can read.

The same policy can become a portable execution envelope: explicit actions, spend ceiling, and a traceable decision. The interface above previews the logic locally.

Try a policy
policy.preview.jsonEXAMPLE
{
  "agent": "research-agent",
  "allow": [
    "market.read",
    "model.infer"
  ],
  "dailyBudgetUsd": 12,
  "onDecision": "append_trace"
}
EXPLICIT BY DEFAULT
04 / THE APPLICATION TOKENUTILITY DESIGN / PLANNED
C·

$CALL belongs
at the boundary.

CALL is the proposed application token for publishing agent policy profiles and allocating shared execution credits. The intended role is operational access inside Callbound, with usage tied to the rules an agent actually follows.

01Policy profile publication
02Shared credit allocation
03Transparent usage receipts
Token issuance, contract addresses, chain, and live credit settlement have not been announced. The workspace is a local product demonstration.