Shaping the Inclusive Future of AI through Education, Community, and Industry Advocacy: H2O.ai

The importance of freedom in the creation and use of artificial intelligence (AI) is multi-faceted and integral to its development. Firstly, fostering innovation relies on an environment that encourages unrestricted exploration of ideas and approaches. Without unnecessary constraints, researchers and developers can push the boundaries of AI capabilities, leading to breakthroughs and advancements. Additionally, an open and collaborative AI ecosystem ensures the incorporation of ethical considerations. This involves principles such as fairness, transparency, and accountability, which are vital for creating AI systems that align with societal values. Moreover, freedom in AI is essential for providing equal access to technology. By avoiding the creation of exclusive, proprietary systems, a more inclusive adoption of AI can occur, empowering diverse communities and preventing the exacerbation of social inequalities. Globally there are various companies delivering excellent services to make AI accessible to everyone but, H2O.ai stands out from the rest. 

Sri Ambati, founder and CEO of H2O.ai driven by a profound personal experience, established H2O.ai with a motivation deeply rooted in leveraging mathematics to combat cancer. Inspired by his mother's diagnosis of breast cancer, Sri's commitment to making a positive impact expanded further during the Delta wave of the COVID-19 pandemic. Leading the initiative o2forindia.org, he facilitated the procurement of O2 concentrators for over 200 public health organizations in India's Tier 2 cities and rural areas, saving numerous lives. Sri's unwavering "AI for Good" motto underscores H2O.ai's mission, emphasizing the responsible and equitable use of artificial intelligence to contribute to the betterment of the world.

In Conversation with Sri Ambati, Founder & CEO of H2O.ai

As the visionary leader in democratizing AI, what are H2O.ai’s key focus areas?

At H2O.ai, our primary focus revolves around democratizing generative AI, ensuring its accessibility and safety for businesses across the spectrum. Simultaneously, we are dedicated to nurturing a robust community of users and contributors. With the emergence of global AI frameworks like NIST, the EU AI Act, and the Biden Executive Order on AI, we recognize the critical need to enlighten and empower AI decision makers and users. Our mission is to provide comprehensive education on evaluating large language models, emphasizing model soundness and validation. This proactive approach aims to establish secure, trustworthy AI, effectively minimizing toxicity, bias, and risks, particularly crucial for regulated industries such as banking, telco, insurance, and healthcare. 

Could you elaborate on how H2O.ai is contributing to the democratization of AI, making it more accessible to a broader audience?

Established in 2012, H2O.ai stands as a trailblazer in the AI landscape, spearheading the movement to democratize generative AI. Our commitment to this mission is evident through the impact of our open source generative AI and Enterprise h2oGPTe, complemented by Document AI and the acclaimed autoML Driverless AI. This transformative suite has empowered over 20,000 global organizations, influencing more than half of the Fortune 500 companies and renowned brands like AT&T, Commonwealth Bank of Australia, PayPal, Chipotle, ADP, Workday, Progressive Insurance, and AES. Beyond our corporate endeavors, H2O.ai is dedicated to making a positive societal impact through the AI for Good program. By supporting nonprofit groups, foundations, and communities, we actively contribute to advancing education, healthcare, and environmental conservation. This includes initiatives to identify areas vulnerable to natural disasters and protect endangered species.

In 2023, H2O.ai took a step further to democratize generative AI by launching free GenAI Masterclass training days at H2O World conferences worldwide, spanning cities like Atlanta, San Francisco, Singapore, and Sydney. This initiative, complemented by over 60 meetups held in the past year, underscores our commitment to evangelizing the power of large language models within the open source community.

What can you tell us about your H2O Generative AI? How does Enterprise h2oGPTe bring value to your clients?

H2O.ai stands out as the sole end-to-end GenAI platform meticulously designed for air-gapped, on-premises, or cloud VPC deployments, providing organizations with unparalleled control over every facet of their stack, from data management to prompt generation. H2O Enterprise h2oGPTe introduces a comprehensive suite of features:

GenAI + Predictive AI: H2O delivers a distinctive fusion of GenAI and automatic machine learning solutions. This integration enables customers to effortlessly develop and deploy use cases by leveraging both capabilities through straightforward API calls.

Cost Control: In the realm of GenAI products, H2O presents an exceptionally cost-efficient solution. Our offering provides unlimited text generation at a fixed cost, presenting an economic alternative that is up to 10 times more cost-effective than pay-by-usage models.

Custom AI Apps: With our low-code Python-based stack, users can swiftly create custom AI apps. Auto-generated code facilitates a rapid start, allowing for easy tailoring of the app to meet specific needs. This flexibility ensures a seamless and adaptable AI application development process for our users.

Could you introduce us to your other products? What are their main features?

H2O.ai proudly offers two flagship solutions: Enterprise h2oGPTe for GenAI and Driverless AI for AutoML. These offerings synergize to empower our customers with full ownership of their models, both generative and predictive. Enterprise h2oGPTe is our cutting-edge platform for generative AI, designed to enhance information retrieval on internal data, host Large Language Models (LLMs) privately, and ensure that your data remains securely within your control. It allows users to integrate any LLM or embedding models, ensuring full scalability with Kubernetes (K8s), and includes comprehensive features such as guardrails, summarization, cost controls, and extensive customization options.

On the other hand, H2O Driverless AI revolutionizes the AutoML process by automating feature engineering, modeling, fine-tuning, and model deployment. It excels at identifying relevant features within a dataset, uncovering feature interactions, handling missing values, deriving new features, and evaluating the importance of each feature. This process transforms features into meaningful values, simplifying their consumption by machine learning algorithms, thereby streamlining the model development lifecycle.

Could you share any success stories where H2O.ai's AI solutions have demonstrated significant impact or innovation?

We're deeply honored to have been a part of numerous customers' AI transformation journeys, witnessing firsthand the transformative power of AI across various sectors. While many success stories are showcased on our YouTube channel and customers' page, AT&T and the Commonwealth Bank of Australia (CBA) stand out for their remarkable achievements. 

Commonwealth Bank of Australia's AI Transformation: CBA has leveraged our solutions to empower over 40 teams, deploying 1000+ models across more than 1 billion data assets. The impact of these deployments spans numerous areas including fraud/compliance, customer 360 views, personalization, retention, and pricing strategies. The results are staggering: 170% engagement uplift in "Next Best Offer" campaigns, a 35% improvement in fraud detection, a direct $500M benefit to customers, and a tenfold productivity increase for data scientists through the use of H2O Driverless AI.

AT&T's AI Journey: AT&T has utilized H2O.ai to tackle a broad spectrum of use cases ranging from marketing and sales to network maintenance, achieving significant milestones:

iPhone Fraud: Transitioning from rules-based systems to AI with H2O AI Cloud, AT&T has significantly reduced iPhone fraud by over 80%, addressing a problem that exceeds $1 billion in the U.S. market alone.

Predictive Maintenance: By predicting battery failures in their 7,000-strong truck fleet, AT&T not only improved operational efficiency but also enhanced customer experience, saving $7 million annually.

Route Optimization: Implementing AI models for route optimization saved AT&T more than $10 million per year by aligning local skills with jobs and minimizing mileage, thereby optimizing fuel consumption.

AI for Good: In collaboration with Argonne National Laboratories, AT&T and H2O.ai developed AI models to predict flooding, validating the efficacy of AI in accelerating research over traditional physics models run on supercomputers.

These narratives underscore the substantial ROI that AI can deliver: dramatically reducing fraud, saving millions in operational costs, and driving efficiencies that directly benefit customers and the environment.

Can you highlight specific industries where H2O.ai sees the most transformative potential through the democratization of AI?

In my experience, I have successfully navigated the intersection of banking, financial services, telecommunications, insurance, and healthcare. H2O.ai, as the sole end-to-end GenAI platform tailored for air-gapped, on-premises, or cloud VPC deployments, uniquely positions itself to cater to regulated industries. This distinctive feature enables compliance and security in sectors where stringent regulations prevail, making H2O.ai an invaluable asset for organizations in these domains.

What is H2O.ai's vision for the continued democratization of AI, and how does the company plan to stay at the forefront of this evolution?

In my role, I am dedicated to H2O.ai's commitment to co-create domain-specific GPTs in collaboration with clients, specifically targeting the distinctive challenges and opportunities within industries like banking, telecommunications, insurance, and healthcare. Our collaborative approach involves close partnerships with customers and industry experts, allowing us to craft solutions finely tuned to the unique requirements of each sector. This strategy not only fosters innovation but also contributes significantly to industry-specific growth, showcasing H2O.ai's ability to deliver tailored solutions that make a lasting impact.

Will H2O.ai be expanding, bringing on any new products that we should be aware of?

We recently introduced our proprietary foundational model, which is now available on Hugging Face with a paper published on Arxiv. While it stands as a leading commercially viable open small model in its category, it’s a  stepping stone for a more robust foundational model in the future.

Considering that many developers lack access to high-end hardware, small models reduce the need to invest in expensive hardware, making them more accessible and cost-effective for developers.

Sri Ambati | Founder & CEO

Sri Ambati stands as a product visionary renowned for assembling world-class teams throughout his career. In 2012, driven by a mission to democratize AI globally, Sri established H2O.ai, sparking a movement that unites top data scientists, physicists, academics, and technologists across over 20,000 organizations worldwide. Actively collaborating with global business leaders, Sri spearheads funding for AI projects addressing profound business, environmental, and societal challenges. As a sought-after speaker and thought leader, Sri has graced prestigious industry events such as Ai4, Money2020, and Red Hat Summit. His expertise has garnered recognition in notable publications, including The Wall Street Journal, CNBC, IDG, and the World Economic Forum, earning him the distinction of being a Datanami Person to Watch.

Prior to founding H2O.ai, Sri co-established Platfora, a big data analytics company subsequently acquired by Workday. Additionally, he served as the Director of Engineering at DataStax and Azul Systems. Sri's academic journey includes sabbaticals focusing on Theoretical Neuroscience at Stanford University and U.C. Berkeley. Holding a master’s degree in Math and Computer Science from the University of Memphis, Sri draws inspiration from his family, including his genius wife and two daughters. His multifaceted career showcases a commitment to advancing AI accessibility globally and finding innovative solutions to complex challenges across various sectors.

“Our mission is to provide comprehensive education on evaluating large language models, emphasizing model soundness and validation. This proactive approach aims to establish secure, trustworthy AI, effectively minimizing toxicity, bias, and risks, particularly crucial for regulated industries such as banking, telco, insurance, and healthcare.”

“Enterprise h2oGPTe is our cutting-edge platform for generative AI, designed to enhance information retrieval on internal data, host Large Language Models (LLMs) privately, and ensure that your data remains securely within your control.”

"We're deeply honored to have been a part of numerous customers' AI transformation journeys, witnessing firsthand the transformative power of AI across various sectors."