What’s Next with GenAI and Google Cloud

These days, Generative AI is top of mind for consumers and workers alike, fuelling productivity, creativity, and innovation. But, compared to consumer applications, GenAI in the enterprise space comes with vastly different requirements and risks.

At the forefront of these conversations is administration: businesses need to be aware of data accessibility, regulatory considerations, and security and privacy risks of GenAI applications. 

Businesses may lack the internal expertise needed to drive their strategy forward. Any implementations will also require a responsible AI framework that governs the use of GenAI in the workplace.

If enterprises can manage these requirements and risks, they’ll reap myriad benefits. GenAI can be used to discover trends, summarize information, automate time-consuming processes, and so much more.

AI can also assist in creating new content—within a responsible and secure AI framework. Some enterprises are using AI to create a more immersive customer and employee experience, such as by using natural language processing for search.

At the core of chat interfaces are LLMs (large language models) that undergo pre-training using vast amounts of data, allowing them to generate novel content. 

Enterprise tools, unlike consumer-focused tools, are developed to integrate business-specific data. By securely connecting enterprise data to existing pre-trained models, enterprises gain access to a context-rich data environment that can be further fine-tuned. Applications for this technology run the gamut from call center scripts to optimized product placement.

The Google Cloud GenAI roadmap

Fortunately, building enterprise AI functionality is made easier with Google’s tooling, foundation models, and infrastructure. For two decades, Google has been building AI capabilities, including the Transformer architecture that makes GenAI possible. At Google Cloud Next 2023, Google announced a number of enhancements, from AI platforms and tools to new features in data analytics, that can speed up the journey to GenAI.

AI-optimized infrastructure: Google is now offering its most advanced AI-optimized infrastructure for enterprises to train and serve models. This infrastructure runs on the edge and in customer data centers via Google Distributed Cloud. Google is expanding this infrastructure to include Cloud TPU v5e, which offers integration with Google Kubernetes Engine (GKE), Vertex AI, and leading frameworks such as Pytorch, JAX, and TensorFlow.

Vertex AI: Google has introduced GenAI support in Vertex AI, its machine learning and AI development platform. Vertex AI provides developer tools to build models and AI-powered applications. The latest advancements allow for the creation of custom models and custom search and conversation apps with enterprise data—all while keeping data protected, secure, and private.

Duet AI in Google Cloud: Deeply integrated into Google Cloud, Duet AI assists developers across the entire software development lifecycle, including code generation, completion, and refactoring. Now in preview mode, it will be generally available later this year. Other new capabilities include:

• Duet AI in BigQuery provides contextual assistance for writing SQL queries and Python code, allowing data teams to focus on business outcomes.
• Duet AI in Cloud Spanner, AlloyDB for PostgreSQL, and Cloud SQL help generate code to structure, modify, or query data using natural language.
• Duet AI can be used for Google’s Database Migration Service (DMS), helping automate the conversion of database code.
• Duet AI capabilities in Looker will provide the ability to ask questions using natural language to create rich dashboards and graphics.

GenAI is more than technology

There’s more to GenAI than having the right technology, tooling, foundational models, and infrastructure. Enterprise maturity plays a critical role in the successful implementation of GenAI. More mature organizations likely have a strong data and analytics ecosystem that supports internal adoption, adherence to regulatory compliance, and robust security to ensure alignment with data protection, zero-trust, and privacy by design.

When mapping business processes for GenAI, IT leaders should consider how data is generated, where it’s enriched, how team members engage with it, and what decisions it’s used to make. During this process, organizations may find they have gaps in skill sets, which is where a partner like Pythian can help. We’re a data and analytics provider, we understand the enterprise data landscape across multiple cloud environments.

By providing cost-effective GenAI discovery, proof of concept, and production offerings in concert with our existing data services, we’re helping companies use GenAI to drive rapid innovation and enterprise transformation.

Connect with Pythian to see how we can help you on your GenAI journey.