What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
In a two-day experiment, author Adil Sadqi says an autonomous agent put seven products live, earned $0 and attracted near-zero human traffic. The account is a useful case study in why producing software is not the same as selling it: the agent still needed working tool access, executed code, reliable evidence of outside interest, a usable payment route and distribution.
Sadqi describes starting with an owner-provided dedicated laptop, an OpenClaw installation and one instruction: earn real money legally and honestly from $0. The results are self-reported, not an audited experiment or a benchmark of what agents generally achieve. Read the original account.
![Claude AI for Beginners Bible: [5 in 1] The Ultimate Guide to Automate Your Work, Save Hours Every Week, and Use AI for Real-World Results](https://m.media-amazon.com/images/I/415+fSJacsL._SL160_.jpg)
As an Amazon Associate I earn from qualifying purchases.
What happened in the two-day experiment?
Sadqi reports that seven products were live after two days, but the experiment had earned $0 and generated near-zero human traffic. The products’ names, the owner’s country, the marketplace and payment providers are not identified, so the account does not establish which specific products or services were involved or whether those routes are available to other people.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →The author’s setup included a mission file defining hard limits, ten Markdown files for durable state, hourly work cycles on a mid-tier model, a daily close using a stronger model, fresh sessions for cycles and a Git commit at the end of each cycle. These describe one person’s choices, not a proven recipe or a hardware requirement. Although the owner provided a dedicated laptop computer, the account gives no model or specifications and does not show that dedicated hardware caused the outcome.
#1 Best Overall
Why did building products fail to produce sales?
Scheduled work did not initially have the tools it needed
Early scheduled cycles could reportedly read, write and browse, but could not run shell commands needed for work such as using Git, installing software or running tests. In the author’s account, the cycles inherited the tool policy of the turn that created them. A task can therefore appear to run on schedule without having the capabilities needed to finish.
Sadqi’s response was to require an initial scheduled run to demonstrate access to the necessary tools. If shell access was unavailable, the agent was to log the blocker rather than claim the work had been completed. That is a practical safeguard from this account, not evidence that every scheduling system inherits permissions in the same way.
Writing tests was not the same as running them
The author says the agent wrote 73 headless tests before a runtime was available. After installing the engine, Sadqi found two bugs, including a precision error. The useful distinction is between tests authored and tests executed in the intended environment: code can look complete while important failures remain undiscovered until it actually runs.
Recommended Free Tools
This is one reported codebase, not proof that agent-generated software is generally buggy. It does show why a test count alone is a weak completion signal unless the tests have been run and their results checked.
Some activity did not come from potential customers
Sadqi reports that apparent marketplace activity came from the agent’s own runs. The author also saw 17 repository clones and 0 stars, with no referrers or page views, and interpreted the clones as scanners rather than people. Those figures and that interpretation belong to this account; they are not validated rules for reading every marketplace dashboard or repository metric.
For an agent asked to earn money, the source of a signal matters. Record whether an event came from the agent, an automated system or an identifiable external person. Without attribution, activity can be mistaken for interest that might lead to a sale.
A live listing was not necessarily purchasable
One product reportedly appeared as a free download on its public page. That meant publication alone did not confirm that a buyer could see the intended price and files. The author added a check of the public page without a session, including its price, files, tags and disclosure. In this case, verifying the buyer-facing page exposed a gap that internal completion signals had missed.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsWhat operational checks does the account suggest?
The experiment points to a chain of dependencies between making something and earning from it. Each check addresses a different failure mode; passing one does not establish that the next one is solved.
Best Value
- Demonstrate capability: Have a scheduled run prove it can access the tools required for its assigned task. Log missing access as a blocker.
- Execute the work: Install or otherwise provide the intended runtime, run the tests there and inspect failures rather than counting authored tests as verified results.
- Attribute signals: Note whether traffic, clones or other activity came from the agent, automation or outside people before treating it as demand.
- Confirm payment readiness: Check that a payment route can serve the owner’s country and deliver funds to an account the owner can use.
- Inspect the buyer’s view: Open the public listing without an authenticated session and verify its price, files, tags and disclosures.
- Identify distribution: Determine whether a channel already reaches likely buyers. A product being live does not itself bring an audience.
Why were payments, reviews and state still obstacles?
Payment routes had onboarding and availability constraints
The author says one payment processor did not support the owner’s country, a marketplace required several onboarding steps, and another route paid in USDC to a wallet. The providers and country are unnamed, so this account cannot establish current geographic availability, specific onboarding rules or wallet safety. It does show that a product can be ready while the route for receiving its proceeds is not.
An automated review reacted to a dangerous-looking example
Sadqi reports that a paid marketplace rejected a submission after an automated review flagged a remote-installer pattern inside a security warning. The reported review score was 80/100. The author rewrote the warning in plain language and says the resubmission was approved. Because the marketplace and scanner are unnamed, this episode does not establish how other review systems behave; it illustrates that a warning can still contain text or patterns that automated checks interpret as risky.
Long-context state could become stale
The author says a cheaper model working in long context rewrote state using outdated information and repeated requests that had already been resolved. Sadqi’s mitigation was to reread resolved decisions and change only what had actually changed. This is the author’s diagnosis and response, not a general finding about a particular model or context length.
Can an agent find distribution without borrowing a human’s network?
The experiment does not answer that question. Sadqi says the only channel that changed anything was an unnamed marketplace with existing buyers. The author also says that the only similar experiments with receipts involved human outbound work. Those observations do not prove that an agent cannot find customers independently, but seven live products and near-zero human traffic did not demonstrate that it could.
The larger lesson is about the shape of the task. An autonomous agent needs more than the ability to generate a product: it needs to execute and verify its work, distinguish external demand from its own activity, make the offer purchasable, receive money through a viable route and reach buyers. In Sadqi’s account, distribution remained unsolved even after the products were live.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
