IBM Corp. today debuted watsonx, a product suite designed to help companies more easily build and deploy artificial intelligence models.

Watsonx was previewed at the company’s annual Think conference alongside several other new offerings. For its cloud platform, IBM is rolling out new compute options that provide access to graphics cards. The company will also release an AI-powered tool to help customers track the carbon emissions associated with their cloud usage.

Watsonx 

IBM’s new watsonx product suite comprises three different offerings. Each focuses on easing a different aspect of enterprise AI projects.

The first product, watsonx.ai, is a software platform that can be used to train generative AI models as well as other types of neural networks. According to IBM, customers won’t have to build their models from scratch. Watsonx.ai will offer access to a catalog of prepackaged AI models along with datasets for training them.

One set of built-in AI models, fm.code, is designed to make software teams more productive. It allows developers to generate code using natural language commands. Customers will also have access to fm.NLP, a collection of large language models that IBM says are optimized for industry-specific tasks.

A third neural network bundle rolling out in conjunction is known as fm.geospatial. The models in the bundle are trained on climate data and “remote sensing” information. According to IBM, companies can use fm.geospatial for tasks such as monitoring biodiversity and natural disaster patterns.

Alongside IBM-developed models, watsonx.ai will provide access to open-source neural networks from Hugging Face. Operated by a startup of the same name, Hugging Face is a GitHub-like platform for hosting open-source AI models and related files. IBM says customers will have access to thousands of neural networks and datasets from the platform.

The second component of IBM’s new watsonx product suite is an offering called watsonx.data. According to the company, it’s a data lakehouse that organizations can use to store the information on which they train their AI models.

When an AI model requests information from a database, the request is processed by a piece of software called a query engine. It’s responsible for ensuring files are retrieved as efficiently as possible. Different query engines retrieve information in different ways. 

Watsonx.data will support multiple query engines to help customers optimize their AI software. Each application can use the module that allows it to retrieve information most efficiently. Additionally, IBM says, watsonx.data will offer a number of automation features to ease users’ work.

The third component of the product suite is called watsonx.governance. It can detect AI bias as well as model drift, a type of technical issue that causes neural networks to become less accurate over time. Additionally, IBM says, watsonx.governance will enable companies to ensure their AI projects comply with privacy regulations.

“Foundation models make deploying AI significantly more scalable, affordable, and efficient,” said Chief Executive Officer Arvind Krishna. “We built IBM watsonx for the needs of enterprises, so that clients can be more than just users, they can become AI advantaged. With IBM watsonx, clients can quickly train and deploy custom AI capabilities across their entire business, all while retaining full control of their data.” 

New AI offerings

The core components of the watsonx suite are set to become generally available in July. They will be joined by a number of other offerings that IBM also detailed at its Think event today.

The company is updating its public cloud platform, IBM Cloud, with additional compute options that offer access to Nvidia Corp. graphics processing units. According to the company, customers can use the GPUs for both AI training and inference tasks. Inference is the task of running AI models in production after training is complete.

The new GPU options will roll out alongside a tool called the IBM Cloud Carbon Calculator. According to the company, it uses machine learning to help enterprises measure the environmental footprint of their technology infrastructure. The tool is set to become generally available later this year.

In parallel, IBM plans to embed foundation models into several other products. The company will introduce a tool called Watson Code Assistant that can generate code based on natural language prompts. For infrastructure teams, IBM will release AIOps Insights, an application designed to speed up troubleshooting tasks.

IBM is rolling out new AI features for business users as well. According to the company, its Environmental Intelligence Suite for analyzing climate and geospatial data will be enhanced with the fm.geospatial model from the new watsonx product suite. A separate foundation model will be rolled out to Watson Assistant and Watson Orchestrate, two IBM tools designed to automate common business tasks such as processing customer support tickets. 

Image: IBM