Whether something was made with AI matters less than whether you were told, and whether the way it was made suits what you need. Softqora asks sellers to label every listing with one of four work modes so that buyers can choose on purpose.
What the four labels say
| Label | Who does what | Think of it as |
|---|---|---|
| Human Built | A person produces the deliverable. AI is limited to routine editor help. | Craft |
| AI Assisted | A person leads and is accountable; AI speeds up drafts and research; the seller reviews everything. | A skilled person with power tools |
| AI Generated + Human Verified | AI produces the first version; a named human checks it against the brief. | Fast output with a quality gate |
| Automated | Software or an agent produces delivery with little or no human involvement per order. | A product, not a service |
Match the label to the risk
Ask what a mistake costs. Where a mistake is cheap to spot and fix, such as a batch of product descriptions or boilerplate code, AI-assisted or AI-generated work with human verification is often a sensible way to get speed. Where a mistake is expensive or hard to see, such as security-sensitive code, legal wording, or anything with your name on it that shapes how people see you, lean towards work where a person takes responsibility end to end.
Judgement and taste favour people
Positioning, brand direction, naming and original creative work depend on judgement about audience and context. AI can generate options, but choosing between them is the work. If that choice is what you are paying for, Human Built or AI Assisted with a clearly senior human lead is usually the better fit.
Repeatable output favours automation
A data clean-up that you will run every week, a template you will reuse, an integration that works the same way every time: these are good candidates for Automated delivery, because the value is in consistency. For these, ask about testing, failure handling and what happens when the input is unusual.
Questions worth asking any seller
- Which parts of this will AI produce, and which parts will you write or decide yourself?
- Who reviews the output before I see it, and what do they check?
- What happens if I find an error after delivery?
- What rights do I receive in the final work? (The law on AI-generated material varies by country, so get the answer in the order.)
What a label does not tell you
A label is the seller's statement about their process. It is not a quality rating. It becomes more reliable when the seller has passed the AI-use check in Softqora Verify, which shows only after the check has actually run. If a label turns out to be wrong, the Dispute Resolution Policy covers refunds and consequences for the seller.