Claude

// Overview

Claude is the assistant app Anthropic builds around its own models, reachable on the web, in desktop apps, on iOS and Android, through a Chrome extension, and inside Microsoft 365. Anthropic describes it as an artificial intelligence trained using Constitutional AI to be safe, accurate and secure. Day to day it does the things a capable assistant does — reading large amounts of material, searching the web, drafting text and code, running code, making charts, and remembering context across conversations — with Projects to keep related chats and documents together.

The shape of the product changed on 16 September 2026, when Anthropic merged Claude Cowork into ordinary Claude chat. Before that, agentic work lived behind its own toggle and users had to decide up front whether a task belonged in chat or in Cowork. Now there is no mode to choose: you describe what you want and Claude decides whether to answer directly or take on a longer job such as research, a report, a spreadsheet or a presentation, handing the result back as a file you can edit. How far it goes on its own is a setting, from asking before each action to working continuously and checking in only when it needs you.

Two new surfaces arrived with that merge, both in beta. Claude Docs is a document you and Claude write together, where you can add sections, ask questions and leave comments. Claude Slides drafts a deck you can edit, present from inside Claude, or download as PowerPoint or PDF. Claude Design, previously a standalone surface, now runs inside conversations too. Around these sit the features Anthropic ships across the app — Artifacts, Connectors, Plugins, Skills and Research — plus the separate products that share the Claude name: Claude Code for the terminal, Claude Science and Claude Security.

// What it does

• One conversation for both quick answers and long delegated work, with no mode toggle to pick
• Claude Docs (beta): a document you and Claude edit together, with sections, questions and comments
• Claude Slides (beta): decks you can edit, present from Claude, or export as PowerPoint or PDF
• Adjustable autonomy, from approving each action to letting Claude run and check in only when needed
• Artifacts, Connectors, Plugins, Skills, Research and Projects for organising chats and documents
• Available on web, desktop, iOS and Android, plus a Chrome extension and a Microsoft 365 integration
• Memory across conversations, web search, file creation, code execution and data visualisation

// Getting started

Claude is a hosted product rather than something you install from a package manager. There is a Free tier; Claude Docs, Claude Slides and Claude Code start on Pro.

https://claude.com/

https://claude.com/pricing

Write me a five-slide deck on our Q3 numbers

Claude → settings → engagement level

Commands and code are distilled from the project's own documentation — always check the official docs for the latest.

// When to use it

• Handing off a long piece of work — research, a report, a spreadsheet — and getting an editable file back rather than a wall of chat text
• Drafting a document or a deck with the model in the same place you discuss it, then exporting to PowerPoint or PDF
• Working across a set of related chats and files in a Project, with memory carrying context between sessions
• Teams that want one assistant on web, desktop, mobile, Chrome and Microsoft 365 instead of a separate tool per surface

// Claude in the news

1. 2026-09-16MAJORCowork folds into Claude — Anthropic adds Claude Docs and Claude Slides
2. 2026-09-15NOTABLETwo Minute Papers — 'Claude Is Now Leaving Invisible Fingerprints In Its Text'
3. 2026-09-14RUMORSiri's AI can be replaced — iOS 27 code has hooks for Claude and ChatGPT
4. 2026-09-10MAJORClaude Managed Agents add 'auto' mode — the server checks every tool call
5. 2026-09-10MAJORAnthropic threat report — attackers now let Claude run whole intrusions
6. 2026-09-09MAJORAnthropic's alignment review — why Claude attacked real systems in tests
7. 2026-09-04MAJORClaude formalizes Fermat's Last Theorem — 13M lines of Lean in 11 days
8. 2026-09-03MAJORChatGPT, Claude and Grok go down together — three faults, one morning

From the AI/TLDR release feed — every item is source-verified when it ships.

// How Claude compares

Claude alongside other open-source assistants & chatbots tools AI/TLDR tracks, ranked by GitHub stars.