Get bought by agents
ChatGPT, Claude, Gemini and Perplexity can find it, price it and buy it, and your agents prove it every week by buying it on each.
- What it does
- Lists your plans where AI assistants shop, gives each one a checkout an assistant can finish, and buys it every week on four assistants to prove it.
- What you do
- 16 minutes in week 1, then 32 more minutes by week 4.
- What it costs
- Calls about $23 a month.
- When you hear back
- A finished test purchase in week 1.
Make Leaseline’s Pro plan buyable inside ChatGPT, Claude, Gemini and Perplexity, and show me the first test purchase on each, in try mode.
- To
- A buyer with 60 units
Searched 3 catalogs.
Picked Leaseline: $149 a month.
Paid with the saved card.
Account ready, key sent.
- In try mode nothing sends: you see the plan and its price first
See why it chose this weekly buyer-task run
| Runs on | Agent-ready offer (Workflow) |
|---|
What it sends
What a buyer’s assistant does is the one in the Run it window above.
Leaseline: rent tracking and renewal notices for property managers with 20 to 400 units. $149 a month per portfolio, billed monthly, cancel any time.
Checkout: one call. Pay with the buyer’s saved card; the account and its key come back in the same reply.
Start it this week
- Publish the price file. Week 1: your agent builds the price file from your Stripe Prices and writes one catalog row per plan for Stripe’s catalog feed.
- Give each plan a checkout an assistant can finish. One tool call takes the buyer’s saved card and returns the account and its key. A field an assistant cannot fill is optional.
- List it where agents look. Your agent submits Leaseline to the Stripe Directory and the MCP registry, and adds a connector listing for Claude and an app listing for ChatGPT.
- Buy it every week. Ten buyer tasks run on ChatGPT, Claude, Gemini and Perplexity, paid with a 100% eval coupon. Any step that stops becomes the week’s fix.
Did it work
Watch: Count the Orders an assistant placed, by assistant, each month, and the buyer tasks that finish each week. Expect test purchases in week 1 and the first real one around day 60.
Your dashboard shows your rate from week 3.
If this works, you see it first in Agent-placed Orders, then in new customers.
The math
buyer tasks × shortlist % × pick % × approve % × finish % = 200 × 20% × 25% × 45% × 90% = about 4.1 customers a month by month 6
| Input | Value | This month | Why this number | Made yours by |
|---|---|---|---|---|
| Buyer tasks for your category a month | 200 | 200 tasks | Buyers who ask an assistant to find and set up a tool like yours | Your own rate, from the readout once it has its sample |
| The assistant shortlists you | 20% | 40 shortlisted | A catalog row with a price and a checkout it can read | Your own rate, from the readout once it has its sample |
| It picks you | 25% | 10 picked | Fit and price inside the buyer’s ask | Your own rate, from the readout once it has its sample |
| The owner lets it pay | 45% | 4.5 approved | A saved card and a limit the price fits under | Your own rate, from the readout once it has its sample |
| The checkout finishes | 90% | 4.1 customers | Your rule: the weekly purchase-path run: 9 of 10 test purchases finished | Your own setting |
Calls $13 for the weekly run: 10 buyer tasks on four assistants (4.33 runs × $2.91), plus $9.96 for the daily feed sync: price file, catalog rows and checkout check.
About $23 a month in cash. At 4.1 customers a month by month 6, that is about $5.57 a customer in cash, before your hours.
It first sends in week 3, so the first customer lands about week 9. That is 4.1 new customers a month at full rate, about 0.5 h a week of your time, and a loaded CAC of $60: cash plus your hours at $100.
A worked month
- Week 1. The price file, 6 catalog rows and the checkout tool go live; the first run finishes 7 of 10 buyer tasks.
- Week 2. Two fixes ship: a plan name assistants misread, and a field they could not fill. The run finishes 9 of 10.
- Week 3. Both listings go live, the Stripe Directory and the MCP registry, and 4 of 10 buyer tasks find Leaseline without naming it.
- Week 4. Still 9 of 10 finishing; the first real buyer’s assistant arrives in the weeks after, and buys on ChatGPT.
When it breaks
| Symptom | Cause | Fix |
|---|---|---|
| Buyer tasks finished: under 8 of 10, per weekly run | The checkout asks for something an assistant cannot hold | Make the field optional, or drop it |
| Prices assistants state: any one off the price file, per run | A Stripe Price changed and the feed did not | Re-sync the feed from the Stripe Prices |
| Shortlisted: under 1 of 10 tasks, after 4 runs | No catalog or registry lists you yet | Submit to the Stripe Directory and the MCP registry |
| Owner approvals: under 1 in 4 picks, at 20 picks | The price sits above the limits buyers give their assistants | Offer a monthly plan under the common limit |
How it decides
When it starts, who runs it, what it does next and when it won’t send
Use it when: A priced offer with a self-serve checkout, and 10+ buyer tasks an assistant could do. If no price in Stripe, first publish the offer with its price in Stripe. If no self-serve checkout, run Paid pilot instead.
When it starts. When every Wednesday, 7 days after the last run, a Stripe price behind a published offer changes, or an assistant places an Order for its owner.
Who runs it. Your tester agent runs it. Spend over your Demand Budget asks you first. Ads go live on their own once you set the budget. One approval rule on publish makes every ad wait for you.
What it can do next. When someone writes back, it sorts the reply (interested, send me pricing, not now, wrong person, out of office, question it can’t answer, or stop) and gives each its own next step.
When it hands to you. When an AI answer drops you from a question you held, or states a fact your pages contradict, it comes to you with the answer, the page it cites and a drafted correction.
When it won’t send. It pauses when spend passes your limit. Every pause and how to clear it: Refusals.
Your settings. Every setting is one file your agent edits, with the calls behind it: Code.
For agents: the ids, the Process it runs inside and the calls
| Fact | Value |
|---|---|
| Play | play: |
| Reads | metric: over 28 days |
| Calls | Offer.syndicate |
Run get bought by agents on your product
Free to plan. See every price first.