Skip to main content
2025

Verdent AI's Transformation: Full-Stack AWS Integration Delivers 40% Cost Savings

Overview

Verdent AI, an enterprise AI coding assistant, empowers developers with intelligent code generation, analytics, and collaboration capabilities, enabling them to benefit from efficient AI-assisted coding services. At its core, Verdent AI utilizes an event-driven architecture to deliver high-performance, real-time communication, offering a suite of nearly 30 capabilities such as code search, intelligent editing, automated testing, and code validation. Currently, Verdent AI's architecture is built on key AWS services such as Amazon Bedrock (AWS Global Regions), Amazon CodeBuild, Amazon CodeDeploy, Amazon RDS, Amazon ElastiCache, Amazon DocumentDB, Amazon EKS, and Amazon CloudWatch.

Missing alt text value

Results

99.9%

Missing alt text value System availability

95%

Missing alt text value Improvement in response speed

40%

Missing alt text value Reduction in intelligent caching

80%

Missing alt text value Improvement in memory efficiency

40%

Missing alt text value Reduction in LLM costs

10x

Missing alt text value Concurrent processing capacity
The AWS team has been generous not only with infrastructure and credits, but also with real conversations, the kind where people sit down with you and help you think through trade-offs. That kind of support is rare, and it's something we don't take for granted. For early-stage teams, momentum is everything. The resources, guidance, and trust from AWS have helped us move faster and with more confidence in our execution. We're building something we believe can make a real impact on engineers, and having AWS in our corner reinforces that belief. Thank you for backing builders at the stage where it matters most.

Xiaochun Liu

COO, Verdent

Opportunities | Using Agentic AI to Overcome Technical Barriers and Build More Stable, Reliable Services

Verdent AI has always been focused on creating enterprise AI coding assistants, such as Verdent for VS Code and Verdent desktop application, aiming to bring the power of intelligent assistants to every developer. However, before implementing Agentic AI, the product was facing significant challenges, including task processing efficiency, quality assurance, and concurrent processing capacity:

  • Inefficient Task Processing: A single AI model struggles to juggle the various stages of complex tasks, such as code generation, validation, and testing, leading to significant inefficiencies.
  • Poor Quality Assurance: Without a dedicated validation system, the code generated by traditional AI assistants is unreliable, which requires significant human oversight and reviews.
  • Limited Concurrent Processing Capacity: Traditional architectures struggle to support large-scale concurrent user requests, resulting in high response latency that significantly increases user churn rates.
  • Unsustainable Costs: When processing extended conversations and large code repositories, the ineffective utilization of multi-cloud LLM resources leads to inflated API costs—averaging $0.5 per query. Also, single-session memory consumption exceeds 500MB, resulting in suboptimal resource efficiency and elevated server expenditures.
  • Suboptimal System Reliability: Single points of failure (SPOFs) render services highly susceptible to outages, with availability merely at 95%—falling short of industry benchmarks.
  • Multi-Model Integration Challenges: Multiple LLM providers are needed for enhanced availability and performance. However, it is hard to schedule and manage without a unified orchestration framework.

To rapidly overcome these barriers, the Verdent AI team required a partner with both a global cloud infrastructure and a comprehensive generative AI ecosystem. With a thorough evaluation, they found that AWS delivers stable model access with superior maturity in managed databases, caching, and document database solutions. Moreover, its highly integrated developer toolchain significantly streamlines their CI/CD pipelines. Meanwhile, AWS provides expedited technical support responses, empowering the Verdent AI team to rapidly resolve critical production environment challenges.

Missing alt text value

Solutions | Enabling Full Lifecycle with AWS, from Infrastructure to AI Services

With the partnership formalized, Verdent AI executed a comprehensive product overhaul leveraging AWS's multi-layered solutions. This transformation optimized resource utilization rates while resolving core challenges in real-time conversational processing and large-scale concurrent request handling:

  • AI Inference Layer: By adopting Amazon Bedrock for a more stable model invocation across global operations, Verdent AI can now process upward of 10,000 concurrent user requests. Simultaneously, intelligent caching strategies have optimized API invocation costs across workload tiers.
  • Data Storage Layer: Verdent AI uses Amazon RDS for saving data such as accounts, configurations, and orders; Amazon ElastiCache for high-performance caching and session state management; and Amazon DocumentDB for storing unstructured data. This approach has significantly boosted data availability, optimized resource efficiency, and allowed for real-time tracking of performance metrics.
  • Containerization Layer: Verdent AI builds a comprehensive system based on Amazon EKS, using Amazon ECR to securely store its services such as Docker images.
  • CI/CD Pipeline: With Amazon CodeBuild, Verdent AI automates the build processes for Go applications and Docker images, while utilizing Amazon CodeDeploy to automate application deployment onto EC2 instances for health checks and verification operations.

Powered by AWS's Agentic AI services, their fully upgraded solution has been applied across diverse scenarios, including software development, open-source project maintenance, and DevOps automation:

  • Software Development: Leveraging Research Agent for in-depth analysis of code architectures, coupled with Verifier Agent to ensure post-refactoring quality, empowers the development team to swiftly accomplish intelligent analysis and refactoring of extensive codebases.
  • Open-Source Project Maintenance: The Verifier Agent automates the linting and test execution, reducing maintainers' workload, with pull request review time significantly dropped from an average of 2 hours to just 15 minutes.
  • DevOps Automation: Automated generation of Terraform/CloudFormation templates, coupled with configuration validity verification via Verifier Agent, accelerates deployment script generation by 10 times.
Missing alt text value
In practical software engineering, the value of AI depends on whether it can support the full lifecycle of a task. Verdent focuses on helping developers move from problem understanding to implementation and verification within a single continuous workflow. AWS Bedrock AgentCore allows us to anchor this workflow in a reliable runtime environment with consistent execution behavior and clear observability. This ensures that every generated change can be reviewed, tested, and integrated with the same standards teams already use, which is essential for maintaining long-term code quality and operational stability.

Zhijie Chen

CEO, Verdent

Business Outcomes | From Single-Model Frameworks to Multi-Agent Intelligent Collaboration Platforms: Dynamically Transforming Services

By using AWS services, Verdent AI has not only accelerated its business innovation, but also amplified its brand influence, effectively ensuring its operational security and compliance:

  • Accelerating Business Innovation: Adopting Amazon Bedrock for its global operations has enabled faster integration of cutting-edge models, accelerated product iteration cycles, and mitigated vendor lock-in risks within multi-cloud LLM architectures. In addition, driven by AWS cloud-native architecture, the multi-agent systems now have achieved a 10x increase in business capacity. Through AWS regionalized deployments, they have optimized global access latency while achieving a 5x acceleration in AI inference capabilities—empowering complex programing tasks and enabling SaaS-centric business models.
  • Optimizing O&M Costs: With intelligent caching and connection pooling optimizations, Verdent AI has cut the cost-per-conversation from $0.5 to $0.3, achieving a 40% reduction in overall LLM invocation costs.
  • Amplifying Brand Influence: Powered by AWS's robust infrastructure, Verdent AI has elevated its system availability from 95% to 99.9%, slashed P95 latency from 5 seconds to just 100ms and scaled concurrent capacity by 10x—significantly enhancing end-user experience while strengthening product credibility.
  • Ensuring Security and Compliance: AWS's security compliance certifications not only fulfill rigorous requirements for high-compliance sectors such as finance and healthcare, but also effectively mitigate multifaceted risks, including data regulation challenges, during the global expansion initiatives.

Looking ahead, Verdent AI intends to leverage Amazon Agentic Core services to deploy Remote MCP Servers, delivering more resilient and stable tooling services for its Agent ecosystem. 

About Verdent AI

The Verdent AI team is dedicated to creating enterprise AI coding assistants such as Verdent for VS Code and Verdent desktop application. Their mission is to enhance global developers' coding efficiency and code quality through AI technology, and deliver intelligent code generation, analysis, and collaboration capabilities—empowering every developer with accessible intelligent programing assistance.