Developers building or debugging agents work through Q&A material scattered across forums, docs and chat logs to quickly find reusable answers or directly generate a working agent.
The public material does not disclose what users did before to solve this problem, so the prior alternative cannot be confirmed.
No evidence in the public record comes from the product itself, its users or its cases, so it is impossible to confirm where agent developers actually get stuck or whether the pain is acute.