We submitted an application for this project and immediately received a test assignment, which we successfully completed, but not without some suspicions, as the client wanted to "make sure we could complete his project." After completing it and receiving the technical assignment, we "shook hands." On the second day, suggestions started coming in like "do it this way" and possibly "we will also need this and that," while the client himself rejected his own technical assignment, claiming that it was supposedly composed by a neural network. That is, in the first version of the indicator that was sent, it corresponded to 80% of everything in the technical assignment, but the client needed additional edits for completely new functions and a design that was not described in the technical assignment initially, because the technical assignment (once again) was composed by a neural network and not properly checked for completeness.
Additionally, outside of the technical assignment, more than half of the volume of the technical assignment was proposed, such settings as alerts of the type, Fibonacci (visual settings, zones, separately for each level) and an interface that was not described, we also noticed phrases like "today we do it this way" and "tomorrow we do it differently" from the client. Thus, the project understandably dragged on due to the constant desire to improve the interface, although this went beyond the scope of the task and did not correspond to the original goals.
The client extended the project twice, as the additional edits we agreed to make required time, but ultimately, after receiving a working indicator with most of the implemented functionality (aside from the interface) - he decided to finish it.