Building Trustworthy AI Agents with Treasure Data’s AI Agent Foundry and Amazon Bedrock

You can’t talk about business today without addressing how artificial intelligence (AI) will impact how you work. AI was a key topic at the recent Gartner Marketing Symposium, where analysts discussed the increasing demands for AI adoption from management, customers, and marketing teams, as well as how AI serves as a strategic lever for growth. 

However, implementing AI in a practical and trustworthy manner isn’t easy to pull off. For many, understanding how to best incorporate AI Agents into work processes is uncharted territory, and doing it right is critical. Workday understands this well. The HR Services provider is facing a collective-action lawsuit that says it built AI Agents that illegally discriminated against older job applicants. 

Building trust with customers has always been an important part of doing business, but it’s now more important than ever as we look at how to build trust in the AI systems we use every day. At Treasure Data, we understand the drive to incorporate AI into business processes to meet the evolving needs of our customers, and we recognize that this must be done in a manner that is well-governed, secure, and trustworthy. So, we thought we would share some insight into how we’re doing exactly that for our customers with the AI Agent Foundry and Amazon Bedrock.

The genesis of trust: Treasure Data AI Agent Foundry

The Treasure Data AI Agent Foundry is an enterprise-grade toolkit for building compliant, auditable, and secure AI Agents in the Treasure Data Customer Data Platform (CDP). The Foundry is a core component of the CDP, not a standalone offering, which means agents adhere to the CDP’s governance framework.

Because the AI Agent Foundry is deeply embedded in the CDP, agents are trained on clean, unified, and permission-controlled customer data (what we call the Diamond Record). Every response, recommendation, or action from an agent is rooted in accuracy, reducing the risk of hallucinations or out-of-context guessing. 

Whether you want to improve personalization or identify opportunities to innovate or develop more efficient processes, AI Agents can provide many benefits. 

The Foundry includes a number of pre-built AI Agents that automate many marketing use cases, including:

• Segment discovery (Audience Agent)
• Campaign planning (Deep Research Agent)
• Content generation (Content Generator Agent)
• Journey optimization (Journey Optimizer Agent) 

You can use any of these agents, customize them to fit your needs, or build custom agents to support your organization’s unique challenges. You can also implement multi-agent orchestration, where an agent can call or delegate tasks to other agents to complete an end-to-end workflow.

Why Amazon Bedrock?

The AI Agent Foundry is built on the Amazon Bedrock infrastructure. Amazon Bedrock is a fully managed service that provides access to a number of foundation models (FMs) from leading AI companies, including Anthropic, DeepSeek, Luma, Meta, Stability AI, and Amazon’s own AI models, along with a broad set of capabilities needed to build generative AI applications. Our decision to use Amazon Bedrock’s infrastructure is based on privacy, security, and scalability.

Bedrock allows us to scale AI applications effortlessly, ensuring our customers have the flexibility they need to grow and adapt. 

With Amazon Bedrock, you can put guardrails in place to build AI Agents safely and responsibly. You retain full control over the data you use to customize a foundation model for your AI Agents, and that data stays under your control. It’s never shared with an FM or used to train their base models. 

Let’s look at some of the reasons Amazon Bedrock is the right AI infrastructure for the AI Agent Foundry. 

Built-in security for trustworthy AI

Amazon Bedrock provides a multi-layered security architecture. From end-to-end data encryption, to virtual private cloud (VPC) and network isolation capabilities, to robust identity and access management control (IAM), and built in threat detection, you get the security and control you need to safely adopt generative AI.

For example, with fine-grained IAM controls, you can precisely define who has access to which data, models, or actions. Resource-based policies add another layer by attaching permissions directly to individual assets. Together, these controls help ensure sensitive data is only accessible to the right people.

To keep your data protected in transit, Bedrock supports private connectivity through Amazon VPC endpoints, which keep traffic off the public internet. Identity-based authentication further strengthens security by verifying every user or system requesting access and enabling full monitoring.

Governance and compliance, built in

The requirement to stay compliant with industry regulations is critical and Bedrock builds governance into its foundation. It aligns with key industry standards and regulatory requirements through automated compliance checks, helping you maintain compliance with less manual effort, and provides automated reasoning tools to act as intelligent guardrails, proactively ensuring compliance and reducing the risk of human error. 

Bedrock also includes two tools that actively monitor and track system activity:

• Amazon CloudWatch provides real-time monitoring, detecting and addressing anomalies and potential breaches.
• AWS CloudTrail provides detailed audit logs documenting all activity ensuring transparency and accountability.

How Amazon Bedrock protects your data

We mentioned the security of your data above but let’s go a bit deeper because it’s essential to understand how secure your data is. First of all, no sensitive data is persistently stored within Bedrock. It also applies data anonymization and data masking to enhance data privacy so even when sensitive information is processed, it’s obscured and protected.

To further strengthen data protection, Bedrock includes secure claims management using Amazon DynamoDB, which helps ensure that all data interactions remain secure and reliable. And it includes automated backup systems that help preserve data integrity and enable quick recovery in the event of a failure or incident. 

Operational benefits that reduce risk

Amazon Bedrock is a fully managed service, which means you don’t need to worry about maintaining security infrastructure or applying security patches. The service does it all for you. 

In addition, real-time security monitoring and proactive threat detection means potential risks are identified and verified before they can impact your operations.  

Building trust with our customers

At Treasure Data, we believe that exceptional experiences and AI hinge on trust. The more you adopt AI into your processes and the more autonomous your AI Agents become, the more critical trust becomes. 

We believe that our partnership with Amazon Bedrock improves both our operational capabilities and our ability to ensure you are implementing trusted AI Agents. Our commitment to you is a reliable and trustworthy foundation that is secure, compliant, and efficient. 

A trusted foundation for AI Agents

As AI becomes more deeply embedded in how businesses operate, building trustworthy, well-governed AI Agents is no longer optional. It’s essential. Treasure Data’s AI Agent Foundry, built on the robust and secure infrastructure of Amazon Bedrock, offers enterprises a powerful yet responsible path forward. 

With built-in governance, compliance, and data protection, it ensures AI Agents not only perform effectively but do so in a way that inspires confidence and trust. For organizations looking to scale AI adoption without sacrificing security, transparency, or control, this is the foundation that makes trustworthy AI a reality.

We invite you to join us in transforming your business with trustworthy AI Agents. Contact us to explore how you can leverage the built-in AI Agents in the AI Agent Foundry or create your own trusted agents built on the secure foundation of Amazon Bedrock.