Skip to main content
please is the financial control plane for AI agents. It is designed to let operators give agents budgets and bounded financial capabilities—not an unrestricted wallet. An agent can request a paid tool or onchain outcome, while identity, policy, budgets, approval authority, simulation, and audit remain outside the language model. The product answers questions such as:
  • Which agent is making the request?
  • Which funding source and policy apply?
  • Is the provider, recipient, asset, protocol, and route allowed?
  • Is the request inside its per-call, task, session, or recurring budget?
  • May it proceed under policy, must a person approve it, or should it be denied?
  • What was purchased, delivered, prepared, submitted, or completed?
The public product is currently a preview. It does not create an agent account, assign a funding source, settle a payment, connect a wallet, approve, sign, submit, broadcast, or move funds.

​
What you can use today

Explore the product preview

Explore fixture-driven policy, funding context, audit, and transaction-review surfaces.

Request a read-only simulation

Use the public Please MCP service from a compatible AI client to discover supported actions and preview one structured action.

​
Authority stays outside the model

AI may interpret a request and explain an outcome. It must not grant itself authority, edit policy, invent arbitrary transaction material, or turn tool output into permission to spend. Deterministic systems own the financial workflow: identity, policy evaluation, budget reservation, typed planning, simulation, approval binding, execution authority, reconciliation, and receipts.

​
Three promises

  1. Bounded capability. Give an agent the narrow financial action it needs, not general access to a wallet.
  2. Deterministic authority. Policy explicitly allows, asks, or denies the exact request.
  3. Readable accountability. Record the agent, purpose, budget, decision, route, settlement, delivery, action, and outcome.

Try the quickstart

Explore the product and run your first read-only preview.

Understand the safety model

Learn what please can—and cannot—do with agent authority, assets, and credentials.