Anthropic

Anthropic is a generative AI company building natural language chatbots with a focus on AI safety research.

The company’s flagship product is a large language model and virtual assistant called “Claude.” Claude can edit, rewrite, summarize, and extract information from large volumes of text, answer questions across a broad range of topics, and automate simple tasks—all via a natural language conversation. Claude is useful in the following areas:

1) Customer service: Handles simple customer queries and workflows and hands off customers to human agents

2) Legal: Extracts information from legal documents

3) Coaching: Guides users through personal growth or career development tasks

4) Search: Allows web search and private search via natural language queries

5) Back-office: Classifies, summarizes, and organizes information from multiple sources

6) Sales: Acts as a sales representative that is capable of answering customer questions while remaining on-brand

Anthropic uses “constitutional training” to tune its Claude model, which is a method that utilizes a combination of reinforcement and supervised learning by leveraging feedback from trained and tuned AI models to ensure Claude behaves in the desired manner. The company aims to develop AIs that can help train new AI models, ensuring they are helpful, honest, and harmless, and aligned with people’s interests. The company claims that it can handle complex multi-step instructions over large amounts of content with 100,000 token windows, compared to OpenAI’s 32,000. 

In September 2023, Anthropic launched a paid version of Claude. The new offering, Claude Pro, was priced at USD 20 (or GBP 18) per month and available in the US and UK. In November 2023, the company upgraded its language model to release Claude 2.1 featuring a more extensive context window, heightened accuracy, and additional extensibility.

In April 2024, the company launched an iOS app and a paid plan targeting businesses, particularly those in regulated sectors such as healthcare, finance, and legal. Named as Team, the plan is priced at USD 30 per user per month, with a minimum of five seats.

Funding and financials

In May 2023, Anthropic raised USD 450 million in a Series C funding round led by Spark Capital. Anthropic planned to use the funding to deploy a new AI model and expand its AI safety research. In August 2023 , Anthropic received a strategic investment of USD 100 million from the South Korean wireless telecommunications operator,  SK Telecom (SKT). As part of the strategic investment, SKT planned to work together with Anthropic to jointly develop a multilingual large language model (LLM) that will enable four Global Telco AI Alliance members, including Deutsche Telekom, e&, and Singtel to offer customized AI developments to their users. In September 2023, Amazon announced plans to invest up to USD 4 billion in Anthropic, initially putting in USD 1.25 billion for a minority stake, with the remaining USD 2.75 billion invested in March 2024. The investment was expected to be used to build a more advanced AI model called "Claude-Next”. In October 2023, Google also agreed to invest up to USD 2 billion in Anthropic. The investment included a USD 500 million upfront cash infusion and an additional USD 1.5 billion to be invested over time to support the development of its AI technology and expand operations.