Verta Inc, the AI Model Management and Operations company, launched the Verta GenAI Workbench, an all-in-one platform to accelerate the Generative AI (GenAI) builder’s journey from idea to product. Geared toward GenAI product builders of all technical levels, the Workbench utilizes a unique combination of automated machine learning and human feedback, enabling builders to rapidly experiment with and refine custom Large Language Model (LLM) solutions for their GenAI applications.
Verta CEO Manasi Vartak says, “The GenAI product development process today is a lot like the Gartner hype cycle: teams start at the peak of inflated expectations based on media excitement or a quick ChatGPT exploration, but rapidly enter a painful valley of disenchantment when they attempt to make the use case work consistently, with high quality, and at reasonable cost. Exiting the valley requires a combination of smart prompting, standing up alternative open-source and proprietary models, data labeling, and infrastructure building, things that most teams don’t have the expertise or tools to do. We built the Verta GenAI Workbench to help teams go from GenAI hype to GenAI productivity as quickly as possible.”
As with traditional machine learning, producing high-quality results from GenAI requires significant trial and error on prompts and models. The Verta GenAI Workbench does this heavy lifting for them. It supports an always up-to-date variety of open-source LLMs like Llama 2, Mistral, and Falcon alongside proprietary models like OpenAI’s GPT models and Google’s PaLM. Similarly, it leverages state-of-the-art prompting techniques and GenAI itself to create “Starter Kits” with strong baseline prompts for a variety of use cases. The result is a single platform that can help builders easily choose and, when needed, switch between different providers and prompts without negatively impacting their GenAI application.
Also Read: Rockset Expands AI Capabilities to Power Billion-Scale Vector Search in the Cloud
An additional differentiator praised by Workbench users is the GenAI Leaderboard, which ranks model and prompt configurations on different dimensions like application-specific quality, cost, and latency, enabling end users to pick the right solution for their specific needs. This also allows non-technical domain experts to steer the results of the application, during product development and also after launch.
“Companies of all sizes are scrambling to integrate GenAI into their businesses; many GenAI experiments are at risk of never coming to fruition due to lack of expertise,” said Mark Rostick, Vice President and Senior Managing Director at Intel Capital. “Verta’s new product simplifies the GenAI app building process significantly, enabling teams to pressure test and build their ideas faster and more efficiently.”
Verta’s commitment to revolutionizing AI technology is not new. Last year, the company was named in Gartner Inc’s “2022 Cool Vendors™ in AI Core Technologies — Scaling AI in the Enterprise.” This accolade underscored Verta’s ability to automate the delivery, monitoring, governance, and scaling of AI models, addressing a crucial need for enterprises to optimize their AI strategies. Building on this legacy, the Verta GenAI Workbench represents the next leap in Verta’s journey of AI innovation.
SOURCE: Businesswire
