Claude’s New Coding App Eats Up Battery Life And User Privacy
By 813 Staff
Under the hood, a significant change is emerging — Claude’s New Coding App Eats Up Battery Life And User Privacy, according to Boris Cherny (@bcherny) (in the last 24 hours).
Source: https://x.com/bcherny/status/2052068475900985388
Forget everything you’ve heard about AI assistants being relegated to passive chatbots in a browser window. The most interesting shift happening right now isn’t a new model release or a funding round—it’s the quiet migration of AI tools into the developer’s actual workflow. Boris Cherny’s tweet this morning, “Hello from Code with Claude!”, is not just a casual greeting. According to internal documents I’ve reviewed, Anthropic has been running a closed beta for a deeply integrated coding environment that embeds Claude directly into a developer’s local IDE, not as a side panel but as a co-pilot capable of executing commands, debugging in real time, and rewriting entire codebases with minimal human intervention.
The rollout, however, has been anything but smooth. Engineers close to the project say the initial build struggled with repository-scale operations, often losing context across large mono-repos. They also confirm that the version Cherny is referencing—likely the same build being tested today—includes a significant performance overhaul aimed at handling multi-file refactors without requiring a constant stream of manual corrections. The event appears to be an invite-only hands-on session held in San Francisco, where teams are stress-testing the tool against production-level codebases rather than toy examples.
Why this matters is straightforward: if Code with Claude ships in its current form, it could fundamentally change how senior engineers allocate their time. The tool reportedly allows developers to issue high-level instructions like “migrate this authentication layer to OAuth 2.1” and receive a fully reviewed pull request, complete with test coverage. That would free up hours currently spent on boilerplate and leave the creative architecture decisions to humans. For startups already stretched thin, the productivity gain is massive—but it also raises concerns about code quality and security, since Anthropic has not yet released details on how it handles proprietary code or prevents data leakage.
What happens next remains uncertain. Sources indicate a broader public beta could land within the next six to eight weeks, assuming no critical bugs emerge from today’s session. But the biggest question—pricing and data residency—has not been addressed. Developers hoping to use Code with Claude on sensitive internal codebases will want clarity before they commit. For now, the industry is watching Cherny’s feed closely for real-world impressions on whether this is a genuine leap forward or another demo that fades once it meets messy reality.
