Zenta AI: Turning Cloud Data into Intelligence with a Claude-Powered Assistant

Zenta AI’s Zenta Pulse platform helps organizations better understand and manage their cloud costs and security posture on Google Cloud.

The platform already gave customers access to detailed billing information, savings opportunities, security findings, and recommendations. But getting from all that data to a simple answer could still take time.

A finance leader might want to know, “Why did our cloud spend increase this month?” An engineer might ask, “What are the most critical security issues we need to address today?”

The answers were available, but finding them often meant navigating dashboards, reviewing multiple findings, or digging through detailed reports.

Zenta saw an opportunity to make that experience much simpler: let customers ask questions in everyday language and get answers based on their own cloud data.

The challenge was making sure those answers were accurate, secure, and reliable enough for a multi-tenant SaaS platform, while keeping AI costs predictable as usage grew.

D3V worked with Zenta to design, build, and launch Ollie, an AI assistant powered by Anthropic Claude and built directly into Zenta Pulse.

Instead of searching through dashboards, customers can simply ask Ollie a question.

When they do, Ollie looks at the relevant billing or security information available within Zenta Pulse and uses that information to create an answer. Responses are grounded in the customer’s actual data rather than relying on the AI model’s general knowledge.

That means a customer can ask questions such as:

“Where are our biggest savings opportunities?”

“Why did our cloud costs increase?”

“What critical security issues should we look at first?”

Ollie can then provide a clear answer and direct the user back to the relevant information in Pulse for additional details.

Common prompts such as Top Savings, Critical Issues, and Explain Trends make it even easier to get started.

The result is a much more conversational way to work with cloud data. Finance and engineering teams can get to the information they need without writing queries or spending time digging through reports.

D3V took Ollie from initial discovery through development, testing, and production rollout, followed by 30 days of post-launch support.

Making Ollie easy to use was only part of the goal. Because it works with customer billing and security information, Zenta and D3V designed the experience around a few important principles.

Grounded in real customer data. Ollie uses current information from Zenta Pulse when answering questions. If the information needed to answer a question isn’t available, Ollie is designed to say so rather than make an assumption.

Customer data stays separated. Zenta Pulse is a multi-tenant platform, so protecting the boundaries between customers was critical. Ollie only receives information the current user is authorized to access.

Read-only by design. Ollie helps customers understand their environment and identify what they should do next, but it does not make changes to customer cloud environments.

Tested before production. Before launch, Zenta and D3V evaluated Ollie against more than 75 real-world questions and tested scenarios designed to verify accuracy, customer-data separation, and resistance to attempts to manipulate the assistant.

Designed for predictable costs. D3V also built controls around how Ollie uses AI so Zenta can manage costs as more customers begin using the assistant.

Ollie runs within Zenta’s existing Google Cloud environment and connects Claude to the information already available through Zenta Pulse.

When a customer asks a question, Ollie identifies the information needed, retrieves the relevant data, and provides that context to Claude. Claude then turns the information into a clear, conversational response.

Behind the scenes, Google Cloud services provide the application infrastructure, data access, security, and monitoring needed to operate Ollie reliably.

The architecture also gives Zenta visibility into how the assistant is being used while maintaining safeguards around sensitive customer information.

For customers, however, the experience is intentionally simple: ask a question and get an answer based on your cloud environment.

The figures below are illustrative placeholders and should be replaced with Zenta’s final production metrics before publication.

• 10x faster answers to common cloud cost and security questions, reducing minutes of dashboard research to seconds
• 65% of active tenants used Ollie within the first 30 days after launch
• 94% accuracy across Zenta’s evaluation set, with reported figures traceable back to retrieved customer data
• Zero cross-tenant data exposures during tenant-isolation and security testing
• Approximately two seconds median time before users begin seeing a response
• Approximately $0.03 average AI cost per conversation, with costs reduced by about 40% through optimization

“Ollie turned our data into answers. Customers can now ask a question in plain English and get a grounded, sourced response in seconds.”
-David, COO, Zenta AI

Ollie gives Zenta a foundation that can grow alongside the Zenta Pulse platform.

Zenta plans to expand Pulse beyond Google Cloud to support additional cloud platforms, including AWS and Azure, as well as AI platforms such as OpenAI and Anthropic.

As those capabilities are added, the same principles behind Ollie (grounded answers, strong customer-data boundaries, and a safe read-only experience) can extend to these new sources.

The goal remains the same: make complex cloud and AI data easier for customers to understand and act on.

D3V Technology Solutions is a Google Cloud Partner and Anthropic Partner helping organizations turn AI ideas into production-ready solutions.

D3V works with customers to design, build, and launch practical AI applications on Google Cloud, with a focus on security, reliability, and measurable business value.