Issue 32: Anthropic and enterprise AI

Last December, we invested in Anthropic at a whopping $15B in valuation. I thought it was very expensive, but compared to its peers and its promise, it was easy to be convinced that perhaps it was cheap. Therefore, I have been watching Anthropic closely. Today I will discuss Anthropic for my readers, while I am still writing my newsletter on Sustainability and AI - have not been able to complete yet.

Introduction

Last week, Anthropic announced the release of?Claude 3.5 Sonnet, a new AI model which has already performed better than most, on the benchmark metrics, and is expected to transform the enterprise AI landscape. Anthropic’s focus on the enterprise market sets it apart from competitors like OpenAI, which has primarily targeted consumers with its?GPT models. Anthropic has successfully established a unique position in the market, by tailoring its products for enterprise applications and pricing them based on factors that are important to enterprises.

Anthropic has an interesting naming convention for its models – Haiku is the smallest, then Sonnet and then Opus. With Claude 3.5 Sonnet they introduced features to understand humor, handle complex workflows and interpret charts and graphs. Following 3.5 Sonnet, we are expecting Opus 3.5 and Haiku 4.0.

Let us look at a table to understand the various models.

While both GPT and Claude models have similar applications, there are specific applications better suited for each.

Claude 3.5 Sonnet seems to have an edge in visual comprehension, code generation, and following specific instructions, making it particularly strong for software development and visual data analysis tasks.

GPT-4o, on the other hand, seems to be more reliable for reasoning and classification tasks, while it has an advantage with superior image generation and voice mode, which could be advantageous in creative and accessibility-focused applications. Again, the focus of enterprise Vs consumer space, seems to be obvious.

The true test is the performance in specific enterprise Vs consumer use cases. For example:

-????????? Claude 3.5 Sonnet can transcribe text from images more accurately – this is super useful in many enterprise applications like digitizing old documents, processing medical records, as well as visual accessibility, while is not as useful to a consumer.

-????????? The large context window is also particularly useful for enterprise projects.

-????????? While Claude 3.5 Sonnet excels in text and image inputs, it did not prioritize speech input or output – as it wasn’t a key business requirement for enterprises

-????????? But that’s not all. Claude 3.5 Sonnet introduced “artifacts”, which at first, seems like an alternate text editor, (for example you can edit an email written by Claude, inside the Claude app, without leaving to go to another editor)…but it is a preview to their vision for an enterprise tool, to securely centralize all documents, knowledge and ongoing work in one shared space – which is a heading towards a full on productivity platform. ?

-????????? While performance is improving, pricing is getting more attractive too - Sonnet 3.5 pricing it at close to a fifth of the price of Opus. Enterprises provided feedback that they care about quality, safety, reliability, speed, and cost and Anthropic is taking care of each of those.

-????????? Lastly, the speed of innovation is particularly interesting to enterprises – as they come up with enterprise optimized updates to one of their three models ever 2-3 months.

Funding History

Anthropic was founded by former OpenAI researchers, led by siblings Dario and Daniela Amodei. The founders left OpenAI due to differences in direction, particularly concerning the ethical and safety aspects of artificial intelligence development. Dario Amodei serves as the CEO, and Daniela Amodei is the President of Anthropic. Their mission focuses on creating safe and reliable AI systems while addressing the broader societal implications of AI technology.

By March 2024, Anthropic had raised a total of $7.6 billion through various rounds of funding, positioning itself as a major player in the AI industry alongside giants like OpenAI, Google and Meta. Anthropic has notable investors like Amazon, Google, Menlo Ventures. Specifically, Amazon has invested a total of $4 billion, while Google has invested $2 billion in the company.

I've compiled this table from many searches (there were numerous discrepancies in the unpaid sources, so there may be some missing rounds).

Benchmark

Here is a detailed view of the performance metrics.

Daniela Amodei, co-founder of Anthropic, can be quoted saying “Claude 3.5 Sonnet is now the most capable, smartest, and cheapest model available on the market today. Out of the standard seven intelligence and capability metrics, Claude 3.5 Sonnet beats all of the competition on six of them and is quite close on the seventh. For standard vision metrics, we’re the top in the leaderboard on four out of five, and a close second on one.”

Certainly quite a feat, and this is one company to watch.

Access:

Users can try Claude 3.5 Sonnet for free at?Claude.ai?and via the Claude iOS app, but with usage limits. Premium subscribers with the Claude Pro or Team plans will have five times as much usage capabilities with their access.

Conclusion

As the AI arms race intensifies, with competitors like OpenAI's GPT-4 and Meta's LLaMA making waves, the launch of Claude 3.5 Sonnet represents a significant milestone in Anthropic's mission to advance the adoption of AI by enterprises, which is will result in enterprises making money and therefore growing.