Today, Anthropic announced a new version of their flagship LLM, Claude 3.5 Sonnet. This new version can use computers.
While AI in RPA tools is not new, the differentiator here is the fact that the model is available right now in public beta. People have already started trying it out.
Casey Newton, as always, provided a great summary about the launch.
Other relevant links:
Even while recording these demos, we encountered some amusing moments. In one, Claude accidentally stopped a long-running screen recording, causing all footage to be lost.
— Anthropic (@AnthropicAI) October 22, 2024
Later, Claude took a break from our coding demo and began to peruse photos of Yellowstone National Park. pic.twitter.com/r6Lrx6XPxZ
RPA tools, and now “AI Agents”—or a combination of both—demonstrate just how inefficient things can be. For the most part, these tools do not rely on structured information such as APIs and databases. In the realm of enterprise software, it is often cheaper to be wasteful with compute resources by parsing the screen, applying a crapload of OCR, and inputting data like an exceptionally fast human.
Unfortunately, I believe this battle was lost long ago.