Claude has been seriously disappointing lately
I’ve been using Claude Max 20x for about six months, mostly Opus, and the results have become extremely frustrating.
My work requires accuracy, specificity, brand voice, citations, and tight control over source material. I’m using it for SEO/content projects, medical-adjacent content, research reports, and high-level brand work where hallucinations or vague assumptions create real problems.
I’ve tried to control for this as much as possible: narrow project sessions, source documents, detailed content maps, strict instructions, brand voice guidance, citation requirements, and checks against provided data.
Even with all of that, I’m constantly correcting the output. It misreads context, makes assumptions, loses specificity, gives generic phrasing, or produces work that sounds right but doesn’t hold up when reviewed. One blog can take two days because I’m rewriting, checking, correcting, and asking why it ignored the instructions or made something up.
At this point, I don’t trust the output enough to use it directly. The model spends half the time apologizing for mistakes instead of producing usable work.
I don’t know if this is model degradation, tuning changes, context issues, or if these platforms just aren’t built for narrow, high-accuracy, brand-specific content workflows yet.
Has anyone else seen Claude get worse lately? Or am I expecting too much from it for this type of work?