Has Claude become noticeably worse over the past 7–10 days, or is it just me?

I’ve been using Claude pretty heavily for the past six months, primarily for my real estate development business. It has gradually become a fairly important part of how I work. I started with Claude Chat, then moved into Cowork and Code. I use it across a pretty wide range of things: Daily management: designing dashboards, to-do systems, tracking whether construction is on schedule, etc. People management: keeping track of different teams, follow-ups, responsibilities and coordination. Brand management: checking whether the brand guidelines are actually being followed across the brochure, website, photography, marketing, etc. Strategy: feeding it very long documents, getting them summarized, understanding the important points, and then asking where they fit into the larger strategic picture of the organization. Construction/project management: using Code to build little internal tools and systems around project tracking. Go-to-market and branding: brainstorming, refining positioning, reviewing work and generally acting as a second brain. For the first several months, I was honestly blown away by how useful it was. It felt like I could give it a fairly complex objective, have a conversation with it, and it would progressively understand what I was trying to achieve. But over the past 7–10 days, something feels noticeably different. The quality of the output has dropped quite significantly for me. I find myself having to give multiple rounds of instructions for things that previously would have taken one or two. More importantly, it feels like Claude is trying to finish the task too quickly rather than understand the task properly. One thing I particularly noticed: earlier, Claude would often stop and ask me several questions before doing the work. Those questions were actually extremely valuable because they helped narrow down what I was trying to achieve. Now it seems much more inclined to just do something immediately — even when the brief is ambiguous — and then I have to spend several rounds correcting it. I’ve tried different models, including the Opus variants available to me, and I’m seeing broadly the same issue. And because I’m using it for fairly complex, interconnected work rather than simple “write me an email” tasks, the difference is becoming quite frustrating. So I’m curious about other heavy Claude users: Have you noticed a deterioration in output quality or reasoning over the past week or two? Or has Claude become more “eager to execute” and less inclined to ask clarifying questions? I’m particularly interested in hearing from people using Claude for Cowork/Code and complex business workflows, rather than just casual prompting. Maybe it’s something about my projects/context getting too large, maybe I’m using it differently, maybe there’s been a change in the models/system prompting — or maybe I’m imagining it. Would be interested to hear if anyone else has experienced the same thing.