Browse categories
Explore
Fiverr Pro
English
$
USD
I will fix one chatbot prompt issue with before and after tests
Independent AI Product Reviewer, Evidence and Reliability
I independently review AI products, assistants and workflows. I built Sigil One and Anubis Review through years of real-world AI testing. I investigate unreliable answers, unsupported claims, instruct...
About this Gig
Does your chatbot keep making the same mistake?
I will correct one agreed prompt or instruction issue and test the change against the original problem and two related cases.
This fits a known issue such as ignoring an approved rule, mishandling missing information, drifting from a required response format, or answering beyond supplied material.
You receive:
- The original prompt and the proposed correction
- Three before-and-after test records
- A short explanation of the change
- Remaining limits and the useful next action
The $75 package covers one issue, one prompt or instruction, up to 1,000 words of supplied prompt/reference material, and up to two additional targeted repeat runs.
Send a failure example, the expected result and where the instruction can be edited. I will confirm that the cause and access fit this package before you order.
Changes are made in an agreed test copy. One in-scope adjustment is included. Backend, retrieval and integration repairs require separate scoping. A proposed change is not called fixed until the agreed checks support it.
Testing application:
Web application
Device:
PC
FAQ
What does the $75 package cover?
One prompt issue in a test copy, up to 1,000 words of supplied text, three before-and-after cases, two additional targeted repeats and one in-scope adjustment. You receive exact change text, results, limits and deployment guidance.
What happens if the checks do not establish a fix?
I report the failed result and agree a next action or marketplace resolution with you. Backend, retrieval and integration repairs need separate scope. Production changes require your explicit authority.
