> ## Documentation Index
> Fetch the complete documentation index at: https://docs.anymorph.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Claude

> Anthropic assistant with optional web retrieval and high-confidence evidence behavior

## Background: provider and product surface

Claude is Anthropic’s general-purpose AI assistant. It can invoke server-side web retrieval when current or external information is needed, but retrieval is a tool rather than the center of the product experience.

## Usage scale and market context

| Metric               | Reported value                          | Context                                              |
| -------------------- | --------------------------------------- | ---------------------------------------------------- |
| MAU/WAU              | Not consistently disclosed by Anthropic | Use third-party traffic estimates only directionally |
| Enterprise/API usage | Significant but not public as MAU       | API, coding, enterprise, and agent workflows matter  |
| Product context      | General assistant + optional retrieval  | Web retrieval is a subset of total Claude usage      |

## Product surface and user experience

* Claude chat or API/tool-use environments can invoke web retrieval.
* Citation metadata may be attached to retrieved evidence.
* Domain allow/block, max\_uses, and user location can shape source pools.
* Answers often synthesize rather than quote sources directly.

## GEO characteristics

| Axis                | Engine-specific behavior                                            | GEO interpretation                                                 |
| ------------------- | ------------------------------------------------------------------- | ------------------------------------------------------------------ |
| Retrieval substrate | Anthropic assistant + optional server-side web retrieval            | Retrieval is a tool, not always the product’s center               |
| Citation UI         | Citation metadata possible                                          | Claim reliability and source quality matter more than source count |
| Query fan-out       | Iterative retrieval for complex comparative/research prompts        | Domain filters/location/max uses can change source pools           |
| Source quality bias | Credible publishers, official sources, research docs likely favored | Verifiable references beat exaggerated marketing                   |
| Measurement note    | Interpret recommendation strength and caveats alongside citations   | Source count alone is insufficient                                 |

### Key cited / influential source types

| Source type                    | Observed pattern                                                                                 | GEO action                                       |
| ------------------------------ | ------------------------------------------------------------------------------------------------ | ------------------------------------------------ |
| Official docs                  | Official product/policy documentation                                                            | Provide accurate claims and limitations          |
| Research/institutional sources | Academic, institutional, government, credible publisher sources                                  | Works as high-confidence evidence                |
| News/publisher sources         | Claude often skews toward publisher and news source types                                        | Earned credible coverage matters                 |
| Structured reference pages     | Definitions, methodology, FAQ, comparison tables                                                 | Logical structure helps retrieval and synthesis  |
| Owned answer surfaces          | homepage inline share 55.1%, pricing/plans 50.3%; product/guide/pricing pages help final answers | Keep official claims precise and non-exaggerated |

## Engine-specific GEO strategy

| Strategy                     | Action                                                           | Metric                          |
| ---------------------------- | ---------------------------------------------------------------- | ------------------------------- |
| Prioritize claim safety      | Use verifiable conservative claims over unsupported superlatives | Claim accuracy                  |
| Build reference pages        | Separate definitions, evidence, limitations, and sources         | Source quality                  |
| Connect to authority sources | Align brand/category info in docs, research, and credible media  | Authority mention               |
| Record tool context          | Track domain filters, location, and max\_uses when testing       | Configuration-specific variance |

## What to measure

* Usage scale and surface-specific exposure
* Source/citation presence and source type
* Brand mention, recommendation, sentiment, and claim accuracy
* Engine-specific failure mode: retrieval, citation, or recommendation

## Sources

* [Anthropic docs: Claude web retrieval tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-search-tool)
* [InfoQ: Anthropic web retrieval coverage](https://www.infoq.com/news/2025/05/anthropic-web-search/)