Google Cloud embeds Gemini across product portfolio
Google Cloud’s Gemini model will power a slew of AI assistant capabilities across its portfolio to bring the benefits of generative AI to developers and enterprise users
Google Cloud is leveraging its Gemini generative artificial intelligence (GenAI) model to power a slew of AI assistants across its portfolio – from Workspace and Chronicle to BigQuery and Vertex AI Studio – in a bid to bring the benefits of GenAI to developers and enterprise users.
For example, with the integration of Gemini in BigQuery, Google Cloud’s data warehouse, organisations can expect the model to have contextual awareness of their business through access to metadata, usage data and semantics.
Other Gemini-powered capabilities in BigQuery include query recommendations, semantic search, low-code visual data pipeline development, and AI-powered recommendations for query performance improvement, error minimisation and cost optimisation.
Additionally, it allows users to create SQL or Python code using natural language prompts and get real-time suggestions while composing queries.
BigQuery and AlloyDB, a managed PostgreSQL compatible database, will also get enhanced vector support, said Brad Calder, vice-president and general manager for Google Cloud platform and technical infrastructure. This will enable organisations to leverage AI where their data is stored, enabling real-time responses for AI applications and building data agents to power their business.
At the same time, Google Cloud is baking Gemini into its Chronicle security operations platform to elevate the skills of security teams and boost their productivity, allowing them to easily detect, investigate and respond to threats using conversational chat with recommended next steps.
“We’re also releasing Gemini in threat intelligence, allowing the use of conversational search to gain faster insight into threat actor behaviour based on Mandiant’s growing repository of threat intelligence,” said Calder, adding that the use of Gemini in the Chronicle security command centre will help organisations evaluate their cloud security posture, find misconfigurations, simulate attacks and receive recommendations for fixes.
Improving efficiency and productivity
For business users, having Gemini in the Workspace collaboration suite can improve efficiency and productivity. For example, PennyMac, a US-based mortgage lender, is using Gemini in Docs, Sheets, Slides and Gmail to accelerate recruiting, hiring and new employee onboarding.
The move by Google Cloud to infuse Gemini across its portfolio follows the recent release of Gemini 1.5 Pro, the company’s multimodal GenAI model that can process up to one million tokens, providing the longest context window of any large-scale foundation model.
That Gemini is being embedded across a broad range of Google Cloud services demonstrates the wide-ranging capabilities of the model – which is trained on a vast dataset comprising code, text, images, audio and video – to address a variety of use cases. Gemini is also seen as key competitive differentiator for Google Cloud as it seeks to gain a leg-up over its hyperscaler rivals in the AI race
The improvements come from the use of Google’s Mixture-of-Experts (MoE) architecture, which increases model capacity without a proportional increase in computation, alleviating the need to fine-tune foundation models or employ retrieval augmented generation to ground model responses in external data.
Google Cloud CEO Thomas Kurian said during a media briefing ahead of Google Cloud Next ’24 that a customer has put an entire code base in Gemini 1.5 Pro because of the model’s long context window to find security vulnerabilities in code.
Gemini 1.5 Pro is available to developers through the Vertex AI platform that organisations can use to deploy and fine-tune AI models, including Google’s Imagen 2, which generates images from text prompts, and third-party models such as Anthropic’s Claude 3 and Meta’s Llama 2.
To help enterprises get the most out of these models, Google Cloud is also rolling out new capabilities to simplify prompt design and help them write good prompts.
With Vertex AI Prompt Management, customers will get a library of prompts used by their teams, including versioning, an option to restore old prompts and AI-generated suggestions to improve prompt performance. They can also compare prompt iterations side by side to assess how small changes impact outputs, along with features such as notes and tagging to boost collaboration.
Finally, they can also use new evaluation tools to assess the quality of models for a set of tasks. A preview feature called Rapid Evaluation lets users evaluate model performance based on metrics such as fluency and instruction-following.
Read more about AI in APAC
SAP’s chief artificial intelligence officer, Philipp Herzig, outlines the company’s approach towards AI and how it is making the technology more accessible to customers.
The Australian government is experimenting with AI use cases in a safe environment while it figures out ways to harness the technology to benefit citizens and businesses.
DBS Bank is building a strong data foundation and upskilling employees on data and artificial intelligence to realise its vision of becoming an AI-fuelled bank.
Boomi CEO talks up the company’s efforts to build up an AI agent architecture, its upcoming AI capabilities set to debut next month, and its footprint in the Asia-Pacific region.
For more robust evaluations, Auto Side by Side (AutoSxS) can be used to compare the performance of two models, including explanations for why one model outperforms another and certainty scores that help users understand the accuracy of an evaluation.
To improve the quality of the answers given by foundation models, Google is also introducing grounding with Google Search.
“We’re bringing the power of the world’s knowledge that Google Search has to ground the answers your model is giving,” said Kurian, noting that this can significantly reduce hallucinations. “We are also introducing grounding with enterprise data, such as your ERP [enterprise resource planning] and CRM [customer relationship management] systems that may have customer data, and Google Cloud databases.”
That Gemini is being embedded across a broad range of Google Cloud services demonstrates the wide-ranging capabilities of the model – which is trained on a vast dataset comprising code, text, images, audio and video – to address a variety of use cases. Gemini is also seen as a key competitive differentiator for Google Cloud as it seeks to gain a leg-up over its hyperscaler rivals in the AI race.
According to a study by Enterprise Strategy Group, 42% of organisations are already using GenAI for a variety of business and IT use cases, and an additional 43% are currently in the planning or consideration phase.
Among survey respondents, expectations were high for targeted use cases, with organisations believing GenAI will most benefit customer service, marketing, software development, IT operations and product development.
