Machine learning operations startup Domino Data Lab Inc. today announced a major update to its artificial intelligence development platform that it says will make it easier for companies to build generative AI models and other applications.

The Fall 2023 release expands the AI Project Hub that was announced in June with various templates and code generating tools to accelerate data science operations, and also adds new access and governance controls.

Domino Data is the creator of a popular MLOps platform used by 20% of the Fortune 100. It encompasses a range of software and tools that make it easier for enterprises to create, fine-tune, deploy and maintain AI models. Its Kubernetes-based platform serves as a single pane of glass for managing everything to do with AI. It can also be used to provision the underlying infrastructure and share models via application programming interfaces.

The startup launched its AI Project Hub initiative earlier this year, saying it will streamline the model development process by providing access to pre-trained foundation models and templates to deliver turnkey AI applications out of the box. It also provides tools for auto-scaling compute clusters such as Spark, Ray, Dask and graphics processing units to enable distributed data preparation and model training at scale.

With today’s update, Domino Data says, the AI Project Hub now provides access to pre-packaged reference projects integrating some of the best known foundation and large language models, to accelerate the development of real-world applications. The new AI Hub Project Templates consist of AI models plus their environments, source code, data, infrastructure and best practices, enabling users to jump start development for a wide range of use cases. Companies can just choose the most suitable template and customize the model for applications such as text generation, Q&A chatbots, sentiment analysis of product reviews, predictive modeling of energy output, fault detection using computer vision and more, the company said.

Domino Data said its launch partners include Nvidia Corp., which has created a customizable Q&A chatbot that can be customized using prompt-tuning techniques with the NeMo Megatron framework. Amazon Web Services Inc. has built a customer service AI model that can be used to summarize product feedback and generate email text, using the Amazon Titan Bedrock service and Anthropic PBC’s foundation models, with LangChain providing additional context.

Another partner is Fiddler AI Inc., which is offering access to Fiddler Auditor, an open-source library for assessing LLMs, while Artefacts AI Inc. is providing a model template for generating structured digital marketing reports with natural language insights. Finally, KSM Technology Partners LLC is offering a model that can be customized to accelerate RNA sequencing for drug development and clinical trials using Nextflow.

According to Domino, it will continue to work with its partners to add more prepackaged templates to the AI Project Hub.

The fall release also introduces the Jupyter AI conversational assistant, which offers a chat interface that can generate entire notebooks from natural language prompts in order to get started quickly with foundational models from Anthropic, AI21 Labs Inc., OpenAI LP and Cohere Inc.

Domino said the new capabilities will prove invaluable for enterprises, because every company is racing to develop generative AI models to boost productivity and enhance their business processes. Those that can get their models up and running in the fastest time possible will gain the biggest advantage, the startup believes.

Enhancing data access and governance

Besides helping customers get up and running faster, Domino Data says it wants to foster a culture of responsible generative AI innovation. To that end, it’s providing immediate, governed access to popular data sources, including Databricks clusters, IBM DB2, Netezza, SAP Hana and numerous others. It makes these data sources available to AI models without moving or transforming and integrating it, thereby reducing the time it takes to customize and fine-tune them.

Finally, Domino Data is integrating a new Data Source Audit LoggingTrails tool with all of its platform’s workflows to ensure governance and validation. According to the startup, it will provide full visibility into every data source its models access, in any environment. It said the tool can be used in conjunction with its existing Domino Model Sentry to provide secure governance and comprehensive monitoring across the entire data pipeline.

Image: Blancoblue/Freepik