# Agent-ready business checklist

From "How to make your business agent-ready" by Sunil Kumar
https://sunilkumar.ai/playbooks/how-to-make-your-business-agent-ready/
Licence: CC BY 4.0. Copy it, edit it, hand it to the team.

The principle: treat the agent as a customer with a different body. It cannot see a PDF, type a one-time code or wait on hold, and it carries its authority, its limits and its money in forms a human never did. Each step gives it one thing it needs.

Score each step 0 to 3:
0 = nothing  1 = a plan or a policy document  2 = built and in use  3 = built, measured and reviewed

Anything under 2 on step 4 (identity) or step 6 (terms) means an agent cannot be trusted at your door, however good the catalogue. A total under 12 means agents are buying from your competitors.

| Step | Score (0-3) | Notes |
|---|---|---|
| 1. Send an agent to buy from yourself | | |
| 2. Publish the offer as data | | |
| 3. Open one door, and watch it | | |
| 4. Verify the agent and bound its authority | | |
| 5. Take an agent's payment | | |
| 6. Write terms for agents, and the unwind | | |
| 7. Give the agent a route to a person | | |
| 8. Measure agent traffic and review it monthly | | |

## 1. Send an agent to buy from yourself

Ask: "Where does an unassisted agent stop when it tries to find our price, buy, change the order and get help?"

- [ ] Four tasks run by an unassisted browsing agent: find, buy, change, get help
- [ ] Every stop recorded with the page, the reason and what the agent did next
- [ ] The same test run against one competitor
- [ ] A named owner for the list and a date to run it again

## 2. Publish the offer as data

Ask: "Can a machine read everything we sell, at what price, with what availability and on what terms, without a person?"

- [ ] Every product or service with price, availability, lead time and conditions as structured data on our own domain
- [ ] The catalogue complete, including the things we would rather discuss on a call
- [ ] Terms attached to the offer, not in a separate document
- [ ] A feed or endpoint we can log, so we know when an agent read it
- [ ] On a platform (Shopify or similar): agent channels on and the catalogue complete

## 3. Open one door, and watch it

Ask: "What are the three things an agent most wants to do with us, and can it do them through one endpoint?"

- [ ] One endpoint, API or MCP server, for the three jobs agents most want (typically price and availability, place or request an order, check its status)
- [ ] Each operation described in plain language with its limits
- [ ] Logging of agent, principal, request and outcome from day one
- [ ] Rate limits of its own, separate from the bot-management rules
- [ ] A route through our bot defences for verified agents

## 4. Verify the agent and bound its authority

Ask: "Can our systems tell a registered agent acting for a customer from a scraper, and stop it the moment consent is withdrawn?"

- [ ] Signed agent requests (Web Bot Auth, Visa Trusted Agent Protocol, Mastercard Agent Pay) verified against a registry
- [ ] Unsigned automation stays blocked
- [ ] One identity per agent, linked to its principal, with a written scope and a cap
- [ ] Revocation that takes effect on the next request
- [ ] A refusal list for agents, enforced at the door, with a reason returned

## 5. Take an agent's payment

Ask: "Which steps in our checkout can only a human complete, and what replaces them for a verified agent?"

- [ ] Scoped tokens, single-use cards or advance mandates accepted
- [ ] Payment provider's agent features enabled (or the conversation started this week)
- [ ] Every human-only step (one-time code, image puzzle, "call to confirm") removed or replaced for verified agents
- [ ] Caps and scopes set by the person in advance; anything outside them escalated, not declined silently
- [ ] A receipt the agent can read and store, in data as well as on a page

## 6. Write terms for agents, and the unwind

Ask: "If an agent buys the wrong thing from us tomorrow, what happens, who pays, and where is that written?"

- [ ] A public terms-for-agents page, linked from the offer and returned by the door
- [ ] Range and prices for agents stated, whether the same as the human channel or not
- [ ] Caps per transaction and per period, above which a person must approve
- [ ] What the agent must present: a verified identity and a principal
- [ ] The unwind for a wrong purchase: window, method, who bears which cost
- [ ] How disputes are raised and answered, by the agent, the person or the agent's provider

## 7. Give the agent a route to a person

Ask: "When an agent is stuck, how does it hand the matter to a person on our side, with the context attached?"

- [ ] A handover channel an agent can use, carrying the transcript and the order identifiers
- [ ] A named owner and a published response time
- [ ] Handover rates tracked per reason, so that recurring reasons become a door or a term
- [ ] The person's answer returned in a form the agent can act on

## 8. Measure agent traffic and review it monthly

Ask: "What share of our visits, orders and tickets came from agents last month, and where did they stop?"

- [ ] Agent share of visits, orders and tickets reported alongside the human channels
- [ ] The step 1 test re-run monthly and the completion score recorded
- [ ] Handover reasons ranked and the top one fixed each month
- [ ] One named owner for the agent channel, with the review in the calendar

## The three numbers to watch

1. Completion of the step 1 test: should rise to five tasks out of five.
2. Share of agent attempts that end in a handover: should fall as reasons become doors and terms.
3. Share of agent attempts that end in a purchase or a completed request: the business case.