> ## Content Index
> Fetch the complete content index at: https://www.testingcatalog.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Anthropic preparing new Agentic Tasks Mode for Claude
- URL: https://www.testingcatalog.com/anthropic-testing-new-agentic-tasks-mode-for-claude/
- Published: 2025-12-16T01:09:05.000Z
- Updated: 2025-12-16T01:11:52.000Z
- Description: Anthropic testing Claude's Agent mode with a new interface for tasks, to introduce new modes for research, analysis, writing, and building.
- Author: Alexey Shabanov
- Tags: Claude News, AI Rumours, Latest AI News

Anthropic is expanding previously discovered [Claude's Agent mode](https://www.testingcatalog.com/anthropic-works-on-agent-mode-for-claude-and-new-avatars/) with a multi-faceted task delegation interface, positioning the assistant as a more autonomous tool for structured workflows. The updated interface introduces a toggle for switching between classic chat and agent modes, allowing users to focus on delegating distinct tasks rather than only having open-ended conversations. In Agent mode, users are greeted with five core sections, Research, Analyze, Write, Build, and Do More, each accessible from the main screen, pointing to specialized capabilities across different use cases.

> BREAKING 🚨: Anthropic will add 5 different starting points to its upcoming Tasks Mode: Research, Analyse, Write, Build, and Do More. Tons of granular controls!  
>  
> A new sidebar for tracking tasks' progress and working with Claude's context has also been added. [pic.twitter.com/DxuKydeh5K](https://t.co/DxuKydeh5K?ref=testingcatalog.com)
> 
> — TestingCatalog News 🗞 (@testingcatalog) [December 16, 2025](https://twitter.com/testingcatalog/status/2000734978305311195?ref%5Fsrc=twsrc%5Etfw&ref=testingcatalog.com)

Selecting a section like Research offers granular control over sources (web or peer-reviewed), effort level, and interaction frequency, suggesting a focus on academic and knowledge-intensive tasks. 

![Claude](https://storage.ghost.io/c/2a/1b/2a1b1782-8506-4d7d-bf53-ad3fb52e2a0f/content/images/2025/12/Claude-12-16-2025_01_49_AM.jpg)

Analyze introduces options for validation, comparison, or forecasting, along with configurable depth and output formats, which could help users in business analysis, reporting, or research validation. 

![Claude](https://storage.ghost.io/c/2a/1b/2a1b1782-8506-4d7d-bf53-ad3fb52e2a0f/content/images/2025/12/Claude-12-16-2025_01_50_AM--1-.jpg)

The Write module provides structured outputs across documents, slides, or spreadsheets, with flexible writing modes and citation settings, addressing varied professional needs. 

![Claude](https://storage.ghost.io/c/2a/1b/2a1b1782-8506-4d7d-bf53-ad3fb52e2a0f/content/images/2025/12/Claude-12-16-2025_01_50_AM--2-.jpg)

The Build area is more visually oriented, enabling selection of themes and layouts, and outputting either interactive artifacts or code.

![Claude](https://storage.ghost.io/c/2a/1b/2a1b1782-8506-4d7d-bf53-ad3fb52e2a0f/content/images/2025/12/Claude-12-16-2025_01_50_AM.jpg)

These functions appear alongside a progress tracker and context manager on the right side of the interface. The progress view will break down each delegated task, while the context manager lists the resources [Claude](https://www.testingcatalog.com/tag/claude/) is using, allowing users to monitor and potentially adjust active inputs. This setup could benefit professionals, students, and teams seeking guided, step-based assistance on complex projects. The company's direction aligns with a broader trend of integrating AI more deeply into productivity and decision-making environments, positioning Claude not just as a chatbot but as an active agent capable of driving workflows from research to final deliverables.