Genspark optimises AI systems with AWS infrastructure

Genspark, a startup that develops general-purpose AI agents and serves more than 20 million users globally, has selected Amazon Web Services (AWS) as its preferred cloud provider. By using AWS infrastructure and generative AI services, the company reports reductions in product development and operational costs, alongside performance improvements in its generative and agentic AI systems.

Founded in 2023, Genspark has created a range of intelligent agent tools, including a deep research agent designed to produce comprehensive results and automate complex research tasks. It has also developed a general-purpose AI agent capable of handling varied tasks without the need for retraining. Its primary product, the Super Agent, provides an integrated workspace supporting research, content creation, and office-related functions.

The company states that AWS autoscaling tools have helped improve the cost efficiency of GPU workloads and simplify operational management. Through the use of Amazon EC2 Spot Instances, Genspark reports cost reductions of 60–70% for GPU resources. The combination of the Cluster Autoscaler for Amazon Elastic Kubernetes Service (Amazon EKS) with EC2 Spot Instances has also supported the provision of required compute capacity at reduced cost.

According to Genspark, AWS’s global infrastructure has contributed to improved access for end users. For example, the company’s Image Studio feature runs its Flux image-generation model on an Amazon EKS cluster using Amazon EC2 G6e instances, enabling faster image generation by drawing on different GPU types.

Developing agentic AI applications requires the integration of model reasoning, planning capabilities, tool use, and prompt orchestration aligned with specific business goals. Genspark uses the Claude family of models on Amazon Bedrock to support agentic AI workflows across products such as Super Agent, AI Slides, AI Sheets, AI Designer, and AI Developer. As a managed service offering access to a range of foundation models, Amazon Bedrock also assists Genspark in reducing training and fine-tuning costs. Its prompt caching feature has helped the company cut inference costs by 72%.

Genspark has taken part in AWS’s “Activate 2.0” programme and participated in the second cohort of the “AWS Startup Accelerator”. These programmes provided technical and resource support that, according to the company, contributed to the validation and rollout of its generative AI products and supported its international expansion.

Justin Liu, co-founder and Chief Architect of Genspark, said that AWS’s technology has supported progress in the company’s agentic AI development and cost optimisation, and that further collaboration is expected. Jason Bennett, Vice President and Global Head of Startups and Venture Capital at AWS, noted that Genspark’s use of AWS infrastructure has enabled cost reductions and supported the scaling of its AI applications.