Ken Yeung

As it seeks to win over enterprise customers, Google is trying to ensure that its AI platform minimizes hallucinations. It's a big deal for organizations, especially those whose executives are already wary about the technology. Google is doubling down on model grounding to reassure them, turning to reputable third-party services Moody's, MSCI, Thomson Reuters and Zoominfo. These four will be available within Vertex AI starting next quarter. They will offer developers qualified data to backstop their model outputs against to ensure responses are factually accurate.

Google's next-generation text-to-image foundation model is coming to the company's Vertex AI platform. Imagen 3 will be available for select customers in preview, offering developers faster image generation, better prompt understanding, a photo-realistic generation of people, and greater text rendering control within an image compared to its predecessor.

Google is adding more third-party models to its AI platform to give enterprise customers more freedom when building apps. Days after integrating Anthropic's Claude 3.5 Sonnet to Vertex AI, Google revealed that it will be adding support for Mistral Small, Mistral Large, and Mistral Codestral this July. Developers will be able to access them through the Vertex AI Model Garden.

Google Cloud is making two variations of its flagship AI model—Gemini 1.5 Flash and Pro—publicly accessible. The former is a small multimodal model with a 1 million context window that tackles narrow high-frequency tasks. It was first introduced in May at Google I/O. The latter, the most powerful version of Google's LLM, debuted in February before being notably upgraded to contain a 2 million context window. That version is now open to all developers.

Google says Gemma 2, its open lightweight model series, will be available to researchers and developers through Vertex AI starting next month. But while it initially only contained a 27-billion parameter member, the company surprised us by also including a 9-billion one.

Figma has introduced native AI features to its design tool to accelerate workflows and enable teams to build high-quality software. Available now as part of a limited beta, Figma AI can generate design drafts using a single prompt, help designers explore different perspectives to express a vision, enable rapid prototyping and iterating, and more. It's part of the company's move to reposition Figma from a design tool to a product development platform to stay competitive against its one-time suitor, Adobe.

Sierra, the customer experience AI startup created by OpenAI board member Bret Taylor and Google AR/VR veteran Clay Bavor, has developed a new benchmark to evaluate the performance of conversational AI agents. Called TAU-bench, agents are tested on completing complex tasks while having multiple exchanges with LLM-simulated users to gather the required information. Early results indicate that AI agents built with simple LLM constructs such as function calling or ReAct don't fare well regarding "relatively simple tasks," highlighting the belief companies need more sophisticated agent architectures.

Bessemer Venture Partners is one of Silicon Valley's most prominent venture firms and one of the earliest to make a bet on SaaS and cloud technologies. Historically, It has parlayed the firm's expertise into annual trend reports such as the State of the Cloud. In 2023, with the rise of ChatGPT, the company declared that language model-powered AI was here to stay. And it was right. For this year's State of the Cloud, Bessemer believes Cloud AI has supplanted the legacy cloud and that the reality about the technology is now outpacing the hype.

Oracle is announcing that its Autonomous Database is now generally available on Microsoft Azure. This is another step by the company to break down the "walled garden of different cloud providers." It's the second Oracle Database to run on Oracle Cloud Infrastructure and is initially available on Microsoft Azure's East U.S. region through Oracle Database@Azure program. There are plans to eventually expand into more data centers in 2024. The technology enables organizations to speed up app development and migrate their on-premises workloads and infrastructure to the cloud.

Ilya Sutskever has revealed what he's working on next after stepping down in May as chief scientist at OpenAI. Along with his OpenAI colleague Daniel Levy and Apple's former AI lead and Cue co-founder Daniel Gross, the trio announced they're working on Safe Superintelligence Inc., a startup designed to build safe superintelligence.

Weeks after AI voice startup ElevenLabs launched its Sound Effects text-to-sound AI offering, the company is releasing an open-source tool to showcase its potential. In "about 15 seconds," this application enables creators to generate sound effect samples for their videos, analyzing the imported clip and providing multiple options.

"In the AI customer support chatbot space, we see more flashy demos than companies that can truly win the trust of an enterprise company," Jesse Zhang, Decagon's chief executive and co-founder, tells VentureBeat. It's a bold statement for a startup just emerging from stealth and entering a market that has become increasingly crowded.