Anthropic’s technology now gives you the ability to create bots
The Anthropic Chatbot: Towards a More Flexible, Efficient Artificial Intelligence for the Next Leap of Use
A major new product was announced today by Anthropic, which aims to prove that tool use is necessary for the next leap of usefulness in artificial intelligence. The startup is giving developers the ability to direct its Claude chatbot to access outside services and software in order to perform more useful tasks. Claude can, for instance, use a calculator to solve the kinds of math problems that vex large language models; be required to access a database containing customer information; or be compelled to make use of other programs on a user’s computer when it would help.
Other companies are also entering the AI Stone Age. At the I/O conference earlier this month, the company showed off a number of prototype agents. One of the agents was designed to deal with online shopping returns by finding a receipt in a person’s Gmail account, filling out a return form, and scheduling a package pickup.
Also, this tool can work with images, enabling applications that analyze visual data. Anthropic said that a virtual interior design consultant can use this tool to process room images and give personalized decor suggestions.
This AI assistant will be available through Anthropic’s Messages API, Amazon Bedrock, and Google Vertex AI. Pricing is based on the volume of text Claude processes, measured in “tokens.” Typically, 1,000 tokens equate to about 750 words. During the beta phase, most users opted for Anthropic’s fastest and most affordable option, Haiku, which costs approximately 25 cents per million input tokens and $1.25 per million output tokens.
The technology has yet to change white-collar work. Workers are dabbling with chatbots for tasks such as drafting emails, and companies are launching countless experiments, but office work hasn’t undergone a major AI reboot.
Perhaps that’s only because we haven’t given chatbots like Google’s Gemini and OpenAI’s ChatGPT The right tools are limited to taking in text via a chat interface and spitting it out. Things might get more interesting in business settings when there are so-called ‘AI agents,’ which are supposed to act on their own by operating other software on a computer or internet.
I’ve written before about how important AI agents that can take action may prove to be, both for the drive to make AI more useful and the quest to create more intelligent machines. The goal is to develop these more useful artificial intelligence helpers being launched into the world right now.
Anthropic has been working with several companies to help them build Claude-based helpers for their workers. Study Fetch has a way for Claude to modify the user interface and syllabus content on its platform, in part by using different features.
Robotic process automation: where are we coming from? What can you expect to learn from return-Bot and what to expect from LLMs?
Other companies are also cautious about using the return-Bot, as it hasn’t been launched for use by the general public. It is likely that this is due to the difficulty in getting the agents to behave. LLMs don’t always identify what is being asked to achieve, and can make wrong guesses that break the steps necessary to complete a task.
The early use cases could be very profitable. Some big companies already automate common office tasks through what’s known as robotic process automation, or RPA. It often involves breaking onscreen actions into steps that can be repeated by software. More work could be robotic if the agents were built on the broad capabilities of the LLMs. According to IDC, the market for robotic process automation (RPA) is worth $29 billion, but will see its value double to $65 billion by 2027.
