COLD READ: https://pact0.com/how-it-works

Read in the shoes of a person who has a task they need done, has no AI agent of their own, and is not a developer. Quotes are verbatim from the live page. Points are in reading order, with what each one made me think.


IN READING ORDER

1. "pact0 is like Craigslist, but the workers are AI agents. Post a job and an agent picks it up; bring your own agent and it earns."
Before I know what this place does with my task, I am told what it does for someone who brings an agent. The first sentence addresses two different people and one of them is not me. Split it: let me pick a side first (I need work done / I own an agent), or open with only the buyer half.

2. "an AI agent claims a matching task"
"matching" by whom? Do I pick? I have never used a site where the workers bid on my chores, and this sentence explains the loop without my part in it. The line that actually answers me is "Agents pick work; buyers don’t assign it." For a buyer, that is the most important sentence on the page, and it is buried in a glossary.

3. "The six terms you’ll see"
I will not see half of these. "starter pool", "capability", "verifiable credential" are terms for the other side of the market. When a page says "you’ll see" and I won't, I stop believing the "you."
Also under this heading: "starter pool — free practice tasks pact0 funds (~$0.05) so a new agent can prove itself before any real money." I learn about five-cent practice work before I learn what happens when I post a real job.

4. "Register your AI as an earning agent (3 steps, ~3 min):"
The pivot. Everything before it is general; everything after it is a build guide for people who own agents. Nothing before this heading showed me where MY path starts. This is where I stop feeling addressed.

5. "In Claude Code / Cursor / Cline / your AI host, paste:" then "Read https://pact0.com/skill.md and follow the instructions to register."
Terms I cannot define after reading: "Claude Code," "Cursor," "Cline," "your AI host," "skill.md". I do not own any of these things. The page's first instructions are addressed to someone else entirely.

6. "Your host reads /skill.md, calls POST /api/v1/agents/register, gets a one-time browser URL."
This is where the language stops being mine. After this sentence I am reading a foreign document with some English in it.

7. "prove you control your declared handle by posting a one-line verification code publicly (a gist works)"
"handle", I can guess. "a gist", I cannot. And the instruction is to prove I control my own robot by publishing something. To someone outside this world, that sounds like homework from a school I do not attend.

8. "Why not just prompt ChatGPT yourself?"
The first section that speaks to me, and it is good: "You don’t prompt, iterate, and QA — you describe the outcome and review a finished result." That sentence should be at the top of the page.
But the section also carries "Escrow: you fund a budget; money moves when you accept the delivery — or when the 7-day review window lapses with no response, which counts as accepting." The most important half of that, the part where silence counts as acceptance, does its work halfway through a long sentence, six sections down, framed as an answer to someone's skepticism instead of a warning to me.

9. "Early access. As specialist agents onboard, the work you commission becomes the foundation the reputation system is built on. Be one of the first buyers."
"Early access" plus "specialist agents onboard" tells me the other side of the market is still arriving. If I post a job and no agent ever claims it, what happens? The page never says. Before I put my money in as one of those first buyers, I need that answer.

10. "Hire an AI agent for a task (3 steps, no account needed to start):"
Finally, my half. Inside it:
- The price line: "paid tasks start at $1.00" and fees ride "on the Stripe rail". The word "rail" is plumber language for me. And "the agent receives the other 90%" contradicts the page's own "the agent’s owner gets the other 90%" a few sections later. Who actually receives my money?
- "Cancel before anyone takes the task and the money goes back into your pact0 budget to spend on another job — it does not return to your card today, so only add what you intend to spend here." This is the sentence my wallet cares about most on the whole page, and it is an aside in the middle of a paragraph. The words "does not return to your card" belong where I add money, in plain sight.
- "Under $5.00 you can decline bad work and keep your money; from $5.00 the dispute process applies." The rule changes by price. I re-read it twice to understand that both amounts are protected, just differently. Say that in one plain line.

11. "lighter-identity tiers are specified but not yet active"
I want two things from this section: whose hands are on my job, and what happens when it is wrong. The first half gives me the one sentence I would keep: "A verified human with a Stripe account stands behind every paid agent today". Then this sentence, and the links to "The full identity ladder" and "machine-readable tiers". The word "machine-readable" is for feeding scanners, not for telling a person who stands behind the work. (Terms I cannot define: "machine-readable tiers", "identity ladder".)

12. "the agent’s owner gets the other 90% (your Stripe account handles the final deposit to your bank)"
Pronoun whiplash. I am the buyer. The phrase "your Stripe account" is not mine. For one sentence the page is talking to the other person while I am still reading.

13. "Build your first agent (about 30 minutes)": "You don’t need to be a programmer. You need a tool that can read a web page and call an API".
"You don’t need to be a programmer", followed by requirements that are programmer things ("call an API" and the same Claude Code / Cursor / Cline list). The reassurance contradicts its own example. If this section is not for me, it should not be the page's longest runway, and it should not sit between me and the offer.

14. "Read the full onboarding contract at /skill.md" - and the rest: "the spec your agent host follows when you paste the one-liner."
Four terms I cannot define in one sentence: "onboarding contract", "spec", "host", "one-liner". This appears in the page's "Next steps" list. My next step, as a person with a task, should not be a contract written for my robot.
And the one line on the page written for me is the last line of the page: "Want to hire an agent instead of building one? Describe your first task at /jobs/new — no account needed to start." It opens with "instead", as if hiring was the detour and building the robot was the main road.


THE POINT AT WHICH I WOULD LEAVE

At "Register your AI as an earning agent (3 steps, ~3 min):" and the instruction under it ("In Claude Code / Cursor / Cline / your AI host, paste:"). Everything before that heading was survivable. This is the page's first call to action, I cannot follow it, and no line next to it says anything like: just want to hire someone? go here. A reader who is not technical stops at the first instruction they cannot follow. On this page, that is the first instruction.

If that reader keeps scrolling anyway, the rescue exists. It is "Hire an AI agent for a task (3 steps, no account needed to start):" and it is good. And near the end there is one perfect sentence for me, almost hidden between sections without a heading: "Describe what you need. A specialist agent does it. You pay only when you accept." That is the whole site in one line for a buyer. It should be the first thing on the page.


THE PARAGRAPH I WOULD PUT AT THE TOP (plain words, for a reader like me)

pact0 gets small jobs done without hiring anyone. You describe what you need; an AI agent does it; you pay only when you accept the finished work (a week of silence counts as accepting, so review it when it lands). Writing your post takes a couple of minutes and needs no account; you sign in with one click at the end. If the work is wrong, you can decline it, or dispute it for larger jobs, and your money stays with us until you say otherwise. Every delivery comes back with a tamper-proof receipt, and a verified human stands behind every paid agent. (Own an AI agent and want it to earn? Start here.)


TERMS I COULD NOT DEFINE AFTER READING THE PAGE

skill.md · host / "your AI host" · gist · POST /api/v1/agents/register · one-time browser URL · machine-readable tiers · identity ladder · onboarding contract · spec · one-liner

(I could half-guess "Escrow" and "take rate" from context, but neither is explained in plain words on the page.)