50% decrease in time required for model upgrades and deployments
Higher uptime with Gemini Enterprise Agent Platform
5x faster response times than other offerings
Moveo.AI, a CX automation company, used Google Cloud to build fast LLMs, enhancing its virtual agents. This boosts Moveo.AI's position in CX, especially for financial services.
Moveo.AI, a CX automation company, used Google Cloud to build fast LLMs, enhancing its virtual agents. This boosts Moveo.AI's position in CX, especially for financial services.
Founded in 2020, Moveo.AI focuses on setting CX in motion by making complex AI technologies more accessible for its customers. Its natural language processing and generative AI solutions infuse AI into customer interactions—but Moveo.AI hit a roadblock when it came to building the right architecture to develop and deploy its LLMs.
The company had taken a multi-could approach to training and serving models, but, fine-tuning third-party models in one cloud and then serving them in another resulted in sluggish response times of 15 seconds or more and significant implementation overhead. With a goal to fine-tune and deploy multi-billion parameter models for its production customers, Moveo.AI evaluated multiple vendors that could support its multi-cloud infrastructure, enable rapid development, and meet data privacy rules, such as the General Data Protection Regulation. Consolidating operations with Google Cloud would empower Moveo.AI to achieve its desired performance, responsiveness, and security.
Our platform empowers businesses to create highly personalized and secure AI agents for exceptional customer experiences. Unlike traditional customer service tools, we focus on driving key business outcomes such as increased revenue and reduced customer churn. By providing pre-trained LLMs fine-tuned for customer interactions, our solution goes beyond basic customer service.
Panos Karagiannis
Co-founder & CEO, Moveo.AI
Accelerating LLM development with Gemini Enterprise Agent Platform
Moveo.AI deployed Google Cloud to benchmark Gemini Enterprise Agent Platform, leveraging A3 VMs powered by NVIDIA's H100 Tensor Core GPUs for its LLM models. Chosen for its speed, access to compute resources, and its global footprint, Agent Platform has supercharged Moveo.AI’s LLM development. The company’s engineers and analysts now use Agent Platform to train its LLMs on all its data sources to craft custom LLMs precisely tuned for CX. This empowers enterprises to develop private, secure, and highly effective CX agents that transcend basic ticket deflection, driving deeper customer engagement, boosting revenue, reducing churn, and building lasting customer loyalty.
The flexibility of Agent Platform allowed the team to maintain granular control over its LLM models while benefiting from the platform's underlying infrastructure. “Gemini Enterprise Agent Platform simplified how we create, deploy, and make these LLMs accessible in a secure way,” says German Attanasio Ruiz, CTO at Moveo.AI. The company’s Co-founder and Chief of AI, George Karagiannis, echoes the sentiment, saying, “We use Agent Platform out of the box with everything managed by Google Cloud. In terms of security, we don’t have to do anything, which is a big plus for us.”
We use Gemini Enterprise Agent Platform out of the box with everything managed by Google Cloud. In terms of security, we don’t have to do anything, which is a big plus for us.
George Karagiannis
Co-founder & Chief of AI, Moveo.AI
To further speed up development and deployment, Moveo.AI uses Dynamic Workload Scheduler, a resource management and capacity scheduling platform that improves Moveo.AI’s access to AI/ML resources. It integrates directly with Agent Platform and scales up or down depending on Moveo.AI’s needs, helping the team optimize spend and improve the experience of workloads such as training and fine-tuning.
Previously, it would take a week to train a variant of Moveo.AI’s LLMs. With Agent Platform, the team trained more than 150 versions in just six months, with each new version introducing new data, improving functionality, and adding new languages and tasks. “This speaks to our team’s abilities overall, but Agent Platform plays a crucial role by giving us access to powerful GPUs, pre-built training environments, and an auto-scaling infrastructure,” the company’s Chief of AI says. “This significantly speeds up our training processes and allows us to iterate much faster.”
As Moveo.AI expands globally and introduces multilinguality to its service, the team will continue to train new LLMs, optimizing for prompt processing and generation speed. They are also expanding on use cases for financial services customers, including support for conversational marketing campaigns, debt collection, and customer service. Moveo.AI is currently benchmarking accelerators in both Google Cloud TPUs (Tensor Processing Units) (Cloud TPU v5e & Cloud TPUv5p) and Google Cloud GPUs - A3 and A2 VMs (powered by NVIDIA H100 Tensor Core GPUs & NVIDIA A100 Tensor Core GPUs) via Dynamic Workload Scheduler.
Powering personalized gen AI experiences
By leveraging infrastructure from Google Cloud and the power of Agent Platform, Moveo.AI aims to accelerate innovation and deliver exceptional conversational agents that drive customer satisfaction and business growth. “If we didn’t have responsive allies at Google Cloud,” Ruiz says, “we would not be able to do this ourselves.”
Moveo.AI is a CX automation company that makes complex AI technology accessible and simple, empowering enterprises to make the most of their data, scale their operations, and offer immersive customer experiences (CX). Headquartered in New York, Moveo.AI has a network of offices in Athens and Sao Paulo to support customers worldwide.
Industry: Technology
Location: United States
Products: Dynamic Workload Scheduler, AI Infrastructure (Cloud TPU v5e, Cloud TPU v5p, A3 VMs (powered by NVIDIA H100 Tensor Core GPUs), A2 VMs (Powered by A100 Tensor Code GPUs), Gemini Enterprise Agent Platform
