Anthropic’s Cloud adds a Prompt Playground to improve your AI apps faster


Prompt engineering has become a popular task in the AI ​​industry over the past year, but it appears that Anthropic is now developing tools to at least partially automate it.

Anthropic on Tuesday released several new features that will help developers build more useful applications with the startup’s language model, Cloud, according to a company blog post. Developers can now use Cloud 3.5 Sonnet to create, test and evaluate prompts, create better inputs using prompt engineering techniques, and improve Cloud’s answers for particular tasks.

Language models are pretty forgiving when you ask them to do certain tasks, but sometimes small changes to the wording of a prompt can lead to big improvements in results. Normally you would have to figure out the wording yourself, or hire a prompt engineer to do it, but this new feature provides instant feedback that can make it easier to find improvements.

These features are housed under a new Evaluate tab in the Anthropic console. The console is the startup’s test kitchen for developers, designed to attract businesses looking to build products with the cloud. One of the features unveiled in May is Anthropic’s built-in prompt generator; it takes a short description of a task and creates a much longer, detailed prompt using Anthropic’s own prompt engineering techniques. While Anthropic’s tools can’t replace prompt engineers, the company said it will help new users and save experienced prompt engineers time.

Within Evaluate, developers can test how effective their AI application’s prompts are in a number of scenarios. Developers can upload real-world examples to the test suite or ask the cloud to create an array of AI-generated test cases. Developers can then compare how effective different prompts are, and rate sample answers on a five-point scale.

A prompt is being fed, generating data to find good and bad answers.
Image Credit: anthropic

In one example from Anthropic’s blog post, a developer found that their application was returning very short answers in many test cases. The developer could make the answers longer by changing one line in their prompt, and apply it to all of their test cases at once. This can save developers a lot of time and effort, especially those with little or no experience with prompt engineering.

Dario Emodei, CEO and co-founder of Anthropic, said in an interview with Google Cloud Next earlier this year that prompt engineering is one of the most important things to achieve widespread adoption of generative AI. “It sounds simple, but 30 minutes with a prompt engineer can often make an application work when it didn’t work before,” Emodei said.


Please enter your comment!
Please enter your name here

Share post:




More like this

Apple Vision Pro adds new immersive video content including movies, TV series

Apple Vision Pro, which was introduced at the company's...

Up to 40% off earbuds, party speakers and more on LG Black Friday in July

As soon as one sale ends, another begins. Such...

How a bad CrowdStrike update crashed the world’s computers

This deep access also increases the likelihood that security...

Last day to save with Disrupt 2024 Deal Days

Today is your last chance to save up to...