Data today is revolutionizing industries, and Windward is at the forefront of transforming the maritime and global trade sector with AI. For over 14 years of intensive data analysis, Windward has delivered predictive insights that make maritime operations safer and more efficient while setting new global standards. So, adopting Amazon Bedrock to enhance its GenAI capabilities was a natural progression, unlocking deeper maritime insights and expanding offerings to its customer base.
However, integrating Bedrock presented the DevOps team with a new challenge:
Enabling teams across the organization to securely and efficiently launch customized and private GenAI environments without bottlenecks.
By leveraging ControlMonkey alongside Bedrock, Windward quickly overcame this hurdle and streamlined the provisioning process with an automated and Terraform-based self-served solution.
The Challenge: Enabling Secure and Scalable GenAI for Every Team
As a data-driven AI company, Windward quickly recognized the transformative potential of recent GenAI developments. Their expertise in processing vast datasets, ranging from maritime logistics to physical vessel security, positioned them to integrate generative AI for deeper insights and predictive capabilities.
When Windward adopted Amazon Bedrock to elevate its GenAI capabilities, leveraging its power within its secure AWS environment was a natural choice. However, implementing Bedrock revealed a significant challenge – making it accessible to teams beyond their proficient DevOps and engineering.
While Bedrock Knowledgebase provided the foundation for building private and tailored GenAI environments that analyze proprietary data, accessing was highly complex for other teams such as:
- RnD / Data Science – Training models on new data sets or libraries
- Sales – Summarizing 500 – page legal documents
- Customer Success – Proactively identifying trends and insights for customers
- Finance – Testing new regulations
The challenges with making Bedrock accessible for other teams:
- The complexity of Amazon Bedrock for non-DevOps
Amazon Bedrock is far from plug-and-play for teams that aren’t proficient with AWS.
Setting up a secure Knowledgebase feature that harnesses Windward’s proprietary data required significant infrastructure management expertise, resources not all teams could access. - Inefficient Processes
Launching a GenAI stack required multiple steps, often involving DevOps teams to handle tickets, prioritize tasks, and manually deploy resources.
This process could take days, stalling progress and productivity. - Cross-Departmental Use Cases
Every team at Windward had diverse requirements so that a one-size-fits-all approach wouldn’t be sufficient.
At the same time, relying on public GenAI tools like OpenAI’s ChatGPT was not an option due to the following restrictions:
- Windward’s highly sensitive maritime data required a fully private and secure solution, making public GenAI tools like ChatGPT unsuitable.
- Sharing proprietary datasets on shared infrastructure posed data exposure risks, regulatory non-compliance, and confidentiality breaches.
Windward prioritized deploying GenAI solutions entirely within its secure AWS environment to maintain control, privacy, and adherence to strict industry regulations.
Windward needed a solution to simplify Bedrock’s deployment for both technical and non-technical users, ensuring that all teams could securely harness its capabilities without compromising data security or operational efficiency.
The Solution: Accessible GenAI Blueprints with Terraform Automation by ControlMonkey
To tackle these challenges, Windward’s DevOps team used a new method, combining Amazon Bedrock and Terraform, to create blueprints that help automate and set up the GenAI infrastructure more efficiently.
Although Terraform was an ideal choice for deploying AWS and GenAI resources, its complexity posed a significant learning curve for non-technical teams. Writing and maintaining Terraform code demanded expertise, and depending on the DevOps team for every deployment hindered innovation and slowed company operations.
Submitting a ticket to the DevOps team to launch and customize a Bedrock Knowledgebase environment, along with the ensuing discussions and other priorities, took a few days.
Windward’s DevOps team tackled this barrier with a unique approach that didn’t require much effort.
They decided to use ControlMonkey’s Terraform CI/CD solution to automate the provisioning of Bedrock and AWS resources, which resulted in a seamless, user-friendly solution for other teams.
- Custom UI with One-Click Infrastructure Creation
Windward’s engineering team created a custom UI for users across the organization.
This user-friendly UI allows anyone to launch a customized Bedrock Knowledgebase Terraform stack tailored to their specific use case without any prior knowledge of Terraform.
Users simply need to input a few variables. With the press of a single button, ControlMonkey’s API automatically plans, applies, and deploys all necessary Amazon Bedrock resources, removing the need for manual intervention. - Enhanced Security and Efficiency
ControlMonkey’s comprehensive automation supports each deployment, ensuring that S3 buckets, OpenSearch indexes, and Bedrock resources are deployed securely behind a VPN with proper IAM roles according to the pre-configured Terraform blueprints. Sensitive data remains private and protected.
Jonathann Zenou, Director of DevSecOps at Windward, said:
“What once took days of back-and-forth communication with DevOps now only takes a few minutes. We removed all frictions by making the process fully automated and self-served, but the real gem is that our data is entirely secure.”
The Transformation: From Days to Minutes, with Zero DevOps Intervention.
By integrating ControlMonkey’s Terraform automation capabilities, Windward changed a lengthy, multi-day process into an immediate operation.
- Faster Deployment: A process that once took days now takes minutes with a simple self-service solution, allowing other teams to deliver faster and focus on innovation rather than the logistics and time of waiting to get a Bedrock environment.
- Enhanced Collaboration: GenAI can be used by non-technical users without relying on the DevOps team. In addition, the DevOps team freed up its engineers to work on more needle-mover tasks.
- Driving Efficiency: By automating infrastructure provisioning, Windward saved precious resources that would have been spent on manual deployments.
- Maintaining Security & Privacy: Sensitive data remains securely within Windward’s AWS environment, meeting the highest privacy standards.
Conclusion: Simplifying GenAI Deployment with Terraform Automation
By combining the foundational capabilities of Amazon Bedrock with the automation and simplicity of ControlMonkey, Windward has created an innovative and self-served model that speeds delivery while maintaining control and security.
Windward’s story demonstrates that even the most complex AI challenges can be simplified with the right tools and vision, enabling teams to unlock new possibilities in record time.ControlMonkey is the most comprehensive Terraform Automation Platform, providing all the necessary tools to automate and govern large-scale cloud infrastructure.
Our Terraform experts would love to jump on a technical call to discuss your use case and show you where ControlMonkey can help.
