CLAUDE LABJP
TEACHERS — Anthropic launches Claude for Teachers, giving verified US K-12 educators free premium access, teaching skills, and curriculum links aligned to standards in all 50 statesADMIN — The Admin API is now in beta for every Claude Enterprise organization: member and invite calls need no beta header, while group and custom-role calls doM365 — The Microsoft 365 connector gains write tools — draft, send, and organize email, manage calendar events and mailbox settings, and create or update files in OneDrive and SharePointMCP — Fixed servers from --mcp-config or .mcp.json ignoring a per-server request_timeout_ms, which left long-running tool calls timing out at the 60s default in fresh sessionsSUBAGENT — A new --forward-subagent-text flag and CLAUDE_CODE_FORWARD_SUBAGENT_TEXT variable include subagent text and thinking in stream-json outputDEADLINE — Opus 4.7 fast mode is removed on July 24, seven days out. speed: "fast" will error, so confirm your move to Opus 4.8TEACHERS — Anthropic launches Claude for Teachers, giving verified US K-12 educators free premium access, teaching skills, and curriculum links aligned to standards in all 50 statesADMIN — The Admin API is now in beta for every Claude Enterprise organization: member and invite calls need no beta header, while group and custom-role calls doM365 — The Microsoft 365 connector gains write tools — draft, send, and organize email, manage calendar events and mailbox settings, and create or update files in OneDrive and SharePointMCP — Fixed servers from --mcp-config or .mcp.json ignoring a per-server request_timeout_ms, which left long-running tool calls timing out at the 60s default in fresh sessionsSUBAGENT — A new --forward-subagent-text flag and CLAUDE_CODE_FORWARD_SUBAGENT_TEXT variable include subagent text and thinking in stream-json outputDEADLINE — Opus 4.7 fast mode is removed on July 24, seven days out. speed: "fast" will error, so confirm your move to Opus 4.8
Articles/Claude Code
Claude Code/2026-05-05Beginner

Claude Code Desktop Got a Major Overhaul: New UI for the Parallel Agents Era

Anthropic redesigned the Claude Code desktop app in April 2026 for the parallel agent era. Explore the new multi-session sidebar, drag-and-drop layout, integrated terminal, Side Chat, and Routines.

Claude Code196desktop app4parallel agentsupdate6AI coding2

I launched Claude Code's desktop app expecting the usual experience, and found the entire interface had changed. The April 14, 2026 redesign isn't a cosmetic refresh — it's a fundamental rethinking of how developers work when multiple AI agents are running simultaneously.

Why Redesign Now?

The previous Claude Code desktop was built around one assumption: you focus on one task at a time. But modern development looks different. "Fix the backend API while another agent writes the tests." "Develop a new feature on one branch while another agent updates documentation." This kind of parallel workflow has become the norm.

Anthropic called this the "parallel agents era" and rebuilt the UI from the ground up to match. The shift is from "question-and-answer tool" to "multi-task orchestrator."

Walking Through the New Features

Multi-Session Sidebar

The most visible change is a session management sidebar on the left. Multiple Claude Code sessions can now be managed within a single window.

Previously, switching between projects meant opening new windows and re-specifying directories. Now it's a single click in the sidebar, with each session's background status visible at a glance. The mental overhead of tracking "where was I in that other window?" drops significantly.

Drag-and-Drop Layout

Sessions can be dragged into split-screen arrangements. Placing production branch fixes on the left and test rewrites on the right becomes effortless. For me, this was the single most welcome change — seeing multiple tasks side by side makes progress management genuinely simpler.

Integrated Terminal and File Editor

A terminal and file editor are now built directly into the desktop app. You can review agent-generated code, make manual edits, and commit — all without leaving the app. The constant tab-switching between Claude Code and VS Code becomes unnecessary.

Side Chat

A way to ask quick questions without interrupting your main workflow. While an agent handles a large task, you can ask a separate Claude instance "how does this library work?" in a side panel. The interruption cost is nearly zero.

Routines (Research Preview)

The most ambitious addition in this release. Routines can run on a schedule, triggered by API calls, or in response to events — and they run on web infrastructure, so your computer doesn't need to stay on.

It's still a "research preview" and has rough edges, but the practical possibilities are already real: automatically summarizing pull request review comments every morning, triggering staging deployments the moment tests pass.

How to Actually Lay Out Two or Three Sessions

Once you can open multiple sessions in the sidebar, the first thing you have to decide is which tasks to run in parallel and which to keep sequential. When you're juggling several projects as an indie developer, it's tempting to launch everything at once — but the combinations that work well share a pattern.

Tasks suited to parallel work are the ones whose files don't overlap. A three-session layout like this, for example, ran reliably for me:

  • Session A: bug fixes on the production branch (only under src/api/)
  • Session B: adding tests on a separate branch (only under tests/)
  • Session C: research, equivalent to Side Chat (not allowed to touch code)

What I avoid is letting multiple agents edit the same file at once. One agent's changes get overwritten by another's, and later you can't tell which version was correct. If you're going to parallelize, it's safest to divide "what each agent may touch" up front by working directory or branch.

The other thing I learned is that the more sessions you add, the more easily you hit rate limits. When three agents run long reasoning passes at the same time, the wait noticeably grows. Keeping heavy generation to one or two sessions and using the rest for review or light questions keeps the overall flow from stalling.

Before You Wire Routines Into Daily Operations

Because Routines run on web infrastructure even when your computer is off, their convenience comes with a catch: they keep running whether you're watching or not. Before I set up any automated run, I always check three things.

First, whether a failure reaches a human. Scheduled runs proceed quietly whether they succeed or fail, so if errors go unnoticed, broken output piles up. Beyond the execution log, I set up a notification path that reaches me when something fails — before turning the automation on.

Second, the fallback model. Claude Code can fall back through up to three models in order, so I specify an alternative for when the primary model doesn't respond, keeping scheduled runs from stalling during peak-time overload. The more an automation runs around the clock, the more that redundancy pays off.

Third, a verification step that judges the output against an independent standard. Rather than applying whatever Routines produced as-is, inserting one lightweight check — does it pass tests, does it fit the expected format — keeps quality from drifting even under automation.

Which Plans Have Access

The redesigned app is available to Pro, Max, Team, and Enterprise plan users, as well as anyone using Claude Code through the Claude API.

The New Role: Orchestrator, Not Typist

What this update signals is a shift in what it means to be a developer. Less time at the keyboard, more time deciding what the agents should do and how they should coordinate. You become the architect of a fleet, not the sole builder.

The new Claude Code desktop is honest about that shift. It's designed at the UI level for developers who manage multiple parallel streams, not just one focused thread. Start by opening two or three sessions side by side — that alone changes how you think about the work.

Share

Thank You for Reading

Claude Lab is ad-free, supported entirely by members like you. We publish practical guides daily with implementation code, benchmarks, and production-ready patterns. If you've found it useful, we'd love to have you on board.

  • Copy-paste ready implementation code
  • New advanced guides published daily
  • $5/mo or $10 for lifetime access
View Membership →

If you found this article helpful, a small tip ($1.50) would mean a lot to us. Your support helps keep this site ad-free and covers server and hosting costs.

Related Articles

Claude Code2026-07-09
Claude Code vs Cursor: The Definitive 2026 Comparison — Choosing the Right AI Coding Tool
A comprehensive comparison of Claude Code and Cursor across pricing, features, accuracy, and workflow. Find the AI coding tool that best fits your development style with our 2026 data-driven guide.
Claude Code2026-07-18
The MCP Server I Declared Vanished from the List — Designing Config for a Namespace You Share with the Vendor
When a vendor reserves an MCP server name, your .mcp.json declaration goes quiet instead of failing. Here is the preflight check and prefix convention I use to turn that silent gap into a loud stop before an unattended run.
Claude Code2026-07-17
The String I Approved Wasn't the String I Read — Testing a Relayed Permission Prompt with Deceptive Characters
I pushed bidi overrides and zero-width characters through my own approval relay. NFKC normalization caught 0%. Here is why, and the implementation that catches 100% with zero false positives.
📚RECOMMENDED BOOKS
Build a Large Language Model (From Scratch)
Sebastian Raschka
LLM Dev
Prompt Engineering for LLMs
Berryman & Ziegler
Prompting
AI Engineering
Chip Huyen
AI Eng
* Contains affiliate links
See all →