CLAUDE

Claude Code Introduces “Agent View”

Anthropic launched “Agent View” in Claude Code. A new way to manage multiple AI coding agents in one place.

Instead of juggling terminal tabs and workflows manually, users can now run agents in parallel, send them to the background, and quickly see which sessions are active, waiting for input, or completed.

If you use Claude Code:

• Run claude agents in the terminal

• Launch multiple agents in parallel

• Use /bg to move sessions into the background

• Switch between active tasks from one interface

The feature is currently available as a Research Preview for Claude Pro, Max, Team, Enterprise, and API users.

PERPLEXITY

Perplexity Launches “Personal Computer” for Mac

Perplexity launched “Personal Computer” in its new Mac app allowing AI agents to work across your local files, Mac apps, browser, and connected tools.

Instead of only answering questions, Perplexity can now help handle longer workflows like organizing files, comparing documents, preparing reports, and managing research across multiple apps and tabs.

To use Personal Computer:

1. Download the new Perplexity Mac app

2. Connect your files and tools

3. Give it tasks in plain English

Examples:

• “Compare these documents and summarize the differences”

• “Organize my project files”

• “Create a report from these PDFs and spreadsheets”

Perplexity says the feature works best on a Mac mini running continuously in the background, with support for approvals and task management across Apple devices.

CLAUDE

Claude Agents Can Now “Dream”

Anthropic shared a new research preview called Dreaming for Claude Managed Agents.

The idea is simple: instead of every AI agent starting from scratch, Claude can review past sessions, spot repeated mistakes, save useful patterns, and update its memory for future tasks.

This means tomorrow’s agents can get better based on what yesterday’s agents learned.

This is currently for developers using Claude Managed Agents, but the bigger idea is worth watching.

Dreaming is designed to help AI agents:

• remember what worked

• avoid repeating the same mistakes

• share learnings across multiple agents

• improve long-running workflows over time

It’s another step toward AI agents that don’t just complete tasks they learn from the work they’ve already done.

MORE AI UPDATES THIS WEEK

Something to read over your cup of coffee ☕️:

1. OpenAI Introduces “Daybreak” for Cyber Defense: OpenAI shared its vision for using AI to help security teams find vulnerabilities, validate fixes, and strengthen software earlier in the development process. The company says AI agents can now assist with code reviews, patch testing, threat modeling, and security analysis directly inside development workflows.

2. Spotify Lets AI Agents Create Personal Podcasts: Spotify introduced a new beta feature that allows AI agents to generate and save personalized podcasts directly into your Spotify library. Users can create custom audio briefings, study guides, travel plans, or daily summaries using tools like OpenClaw, Claude Code, and OpenAI Codex.

3. OpenAI Adds “Goals” for Long-Running AI Tasks: OpenAI introduced a new experimental /goal feature in Codex that allows AI agents to work independently toward a long-term objective for hours at a time. Developers can set a clear end goal, validation checks, and stopping conditions, allowing Codex to continuously iterate on coding tasks, migrations, prototypes, and experiments without needing constant supervision.

4. GPT-5.5 Instant Becomes the New Default ChatGPT Model: OpenAI rolled out GPT-5.5 Instant as the new default ChatGPT model, bringing better reasoning, fewer hallucinations, and more personalized responses. The update also introduces “memory sources,” allowing users to see which past chats, files, or saved memories influenced a response and manage that context more transparently.

5. Microsoft Expands Copilot Cowork Across Skills, Apps, and Mobile: Microsoft introduced new updates to Copilot Cowork, including reusable “Skills,” deeper integrations with tools like Power BI and Dynamics 365, and support for iOS and Android. The company says the goal is to move AI beyond chat and into real task execution across everyday workflows and business systems.

6. OpenAI Expands Advertising Inside ChatGPT: OpenAI introduced new ways for businesses to run ads in ChatGPT, including a self-serve Ads Manager and cost-per-click (CPC) bidding. The company says ads will remain separate from ChatGPT responses, while advertisers can now better track campaign performance without accessing personal conversations or user data.

7. Anthropic Launches Enterprise AI Services Venture: Anthropic announced a new AI services company backed by Blackstone, Goldman Sachs, and Hellman & Friedman to help businesses integrate Claude into everyday operations. The venture will pair Anthropic engineers with enterprise teams to build custom AI workflows across industries like healthcare, manufacturing, and finance.

8. How AI Agent Memory Actually Works: A new deep dive breaks down how AI agents “remember” information across conversations and workflows. The guide explains concepts like context windows, long-term memory, vector embeddings, retrieval systems, and multi-agent memory showing how modern AI assistants are evolving from one-off chatbots into systems that can retain knowledge and improve over time.

TWEETS OF THE WEEK

MORE AI IN ACTION

DISCOVER THESE AI TOOLS AND APPS

1. Jotform Claude App - Build Jotform forms directly in Claude using simple prompts.

2. TOFree AI SEO Auditor - See your page through the eyes of ChatGPT, Claude, and Perplexity. Get a 0–100 visibility score and a copy-paste prompt your coding agent can run to fix what’s broken.

3. Wispr Flow - The voice-to-text AI that turns speech into clear, polished writing in every app.

4. Hopper - The first agentic development environment for the mainframe.

5. Vexilo - An interactive handbook for Claude Code — 31 agents, 92 commands, and 121 skills, organized so you can find what you need fast. This tour takes about 30 seconds. Let's start with the one step that matters most.

6. EmailTemple - Describe what you want to send. Get a high-converting, dark-mode-safe template, exported to your ESP in seconds.

   Generate your own

7. Framer - Framer is the site builder trusted by leading startups and Fortune 500 companies. Build fast and scale more easily with an integrated CMS, analytics, localization, and SEO.

8. Pixcode - Pixcode turns your machine or server into a live workspace for AI coding CLIs. It keeps chat, files, shell, Git/local changes, orchestration, TaskMaster planning, Telegram, notifications, MCP, plugins, and API automation in one browser or desktop app.

9. Khaos Brain - is a “brain-like” experience system for AI agents: it organizes task experience, predictive models, user preferences, runtime lessons, and shareable skills into a card library that is observable, maintainable, and Git-versionable.

10. MY AI Agent - A Slack-style app where you can hire as many AIs as you want. DM one, build a team with several. Delegate work like to coworkers.