Claude Cowork and chat are now one Claude
Anthropic’s move to merge Claude Chat and Cowork into a single unified “Claude” interface is drawing mixed reactions from users. Many welcome the simplification for non-technical people who were confused by multiple modes, but power users worry about losing fine-grained control over memory, tools, costs, and distinct reasoning vs. coding workflows. The debate also surfaces broader concerns about chat as the dominant UX for complex work, comparisons with OpenAI’s value proposition, and a desire for more flexible, model-agnostic and locally hosted alternatives.
Overall reaction to merging Chat & Cowork
- Many see unifying modes as simplifying a previously confusing experience, especially for non‑technical users who didn’t know when to choose chat vs cowork vs code.
- Others strongly dislike losing an explicit “just chat” mode and fear the app will become more opaque and harder to control.
- Several users say they’ll now avoid the web/desktop UI entirely and stick to CLI/IDE or other tools.
UX, modes, and product design
- Prior UX was described as “jagged”: different surfaces (chat, cowork, cloud/local code) behaved inconsistently, sometimes blocking common tasks.
- Some argue one unified entry point is inevitable and good for 95% of users; others insist “one size” forces power users to adapt unnecessarily.
- There are repeated calls for clearer task-based structures (projects, tasks, workflows) instead of long linear chats, especially for multi-day or recurring work.
Memory, privacy, and safety
- Users who kept memory off valued the guarantee that old context wouldn’t leak into new conversations; they worry unified modes plus memory may surface unwanted past details in front of others.
- There are concerns about stronger capabilities (local files, autonomous work) being on by default and removing a simple “low-risk” chat mode.
- Healthcare users say agentic “cowork”-style behavior is unacceptable in regulated environments.
Comparisons with other tools
- Several commenters are switching to or preferring OpenAI (chat + Work/Codex), Codex/Astra, or local models, citing better value, fewer refusals, and clearer separation of quotas.
- Some want a model-agnostic “chat shell” with history, projects, memory, and good search without self-hosting.
Chat as primary interface
- Many criticize chat as a tedious, linear interface that doesn’t scale for complex work, branching explorations, or long-term projects.
- Others defend chat as the simplest universal UI while capabilities and best practices are still evolving.
- There’s interest in alternatives: tree-structured conversations, task/workflow systems, DSL-like UIs, and agent orchestrators that hide most of the chat.
Quality and trust issues
- Users report large differences between “chat” and “work/coding” harnesses in reasoning behavior.
- There’s skepticism toward marketing demos (vacation planning, slide decks) and frustration with people presenting unchecked LLM-generated content.