Lightning AI Inc., the company behind the popular artificial intelligence development framework PyTorch Lightning, is evolving its offerings with the launch of Lightning AI Studios with a goal to simplify the way enterprises can build and deploy AI-infused applications.

Announced today, the new offering delivers a more comprehensive, intuitive and fully integrated experience that enables tasks such as data preparation, model development, distributed training and AI product development to be done in one, easy-to-use environment, the startup said.

With today’s launch, Lightning AI said, it’s offering a tonic to the complex reality of AI product development, which typically involves using multiple tools and disparate platforms to piece everything together. By bundling everything required to do this in a single platform, Lightning AI says it can completely re-engineer AI workflows, with built-in infrastructure management and native integrations shaving months off development times.

The Lighting AI Studios platform is integrated with PyTorch Lightning, a lightweight yet high-performance and open-source framework for training AI models that has racked up more than 91 million downloads globally. The standalone PyTorch Lightning framework already simplifies AI development by automating many of the manual tasks involved in such work.

For instance, when building neural networks, developers must also set up and manage the infrastructure those networks run on. This is a time-consuming process that PyTorch Lightning simplifies through automation.

Another task that PyTorch Lightning simplifies is AI training. It does so using a combination of large numbers of processors and tools that make it easier for developers to evaluate the speed of their neural networks and troubleshoot any issues that come up. Another benefit of PyTorch Lightning is that it reduces the amount of custom code developers must write to use multiple types of chips, such as graphics processing units and central processing units, in their machine learning training projects.

With Lightning AI Studios, developers get additional tools that can integrate the models developed with PyTorch Lightning into fully fledged and production-ready applications.

One of the main benefits of Lightning AI Studios is that there’s no setup process involved. It gives developers everything they need to get started building AI-based applications through a single portal. In doing this, the platform can accelerate speed-to-deployment while meaningfully reducing the costs for both individual developers and enterprise customers, the startup claims.

It also provides a degree of flexibility, with users able to build their AI products on Lightning AI’s default Amazon Web Services cloud, on their own cloud account or inside an on-premises data center. Costs are reported in real time, and users can scale cloud resources up or down to ensure they stay within their budgets.

The startup reeled off a list of benefits, with prebuilt templates to jumpstart AI model development, including the ability to scale up by switching from central processing unit-based virtual machines to more powerful graphics processing units-based VMs in a single click. Its platform also facilitates large language model training from scratch, with data prep capabilities allowing users to customize their datasets first. There are LLM evaluation tools too, as well as application programming interfaces for multimodal image and language chat applications.

With these capabilities, developers can use Lightning AI Studios to fine-tune a range of LLMs, such as Llama 2, Code Llama and Mistral 7B, on their own proprietary data, or alternatively design and train their own LLMs. Once they’re ready, these LLMs can then be fine-tuned for specific workloads and infused with almost any kind of software application., the startup said.

According to Lightning AI, the new offering has seen rapid adoption by early adopters, including individual developers, startups, mid-sized businesses and even some Fortune 100 companies. It’s being used to make a wide range of AI applications in areas such as drug discovery, clinical trials, fraud prevention and risk analysis.

Sam Wolk, a graduate student at MIT’s Sustainable Design Lab, said he has used Lightning AI Studios to build an AI application that can reduce carbon emissions in construction projects. He explained that his AI model simulates multiple construction scenarios to identify the most sustainable method.

“I don’t need to be concerned about environment management, images or dependency management and the Lightning framework handles all the boilerplate work,” Wolk said of the new platform. “This has resulted in a model that only took us six hours to develop with Lightning AI Studios that had previously taken us 36.”

Lightning AI Studios is being made available to individual students, researchers and hobbyists, while there are three subscription tiers for companies. They include a Pro version for engineers, researchers and scientists, a Teams version for startups and larger research teams, and an Enterprise version for the largest organizations.

Image: Lightning AI