Zoe Subscription Intelligence – Zone Control Center

Summary: Use this article to learn how to use the Subscription Intelligence Zoe AI Agent. Article includes best practices and use cases.

Note: Zone Control Center and Zoe mimic each user's NetSuite and Zone permissions. If users have limited access to NetSuite and Zone solution data, those permissions will be automatically applied within Zone Control Center. Therefore customers will not be able to query or see data that they do not normally have access to view.

What is Zoe Subscription Intelligence?

Zoe by Zone is our agentic orchestration layer, built directly into NetSuite. Subscription Intelligence is a Zoe AI agent that can pull data from any NetSuite and ZoneBilling custom record that the current user has permission to access. This includes customer data, subscription data, charge data, usage data, ZoneBilling custom records, and more.

However, Subscription Intelligence is limited by each customer’s role and permissions. For example, if the customer has a limited view within NetSuite and ZoneBilling, then Subscription Intelligence is only able to use the information available to that customer given their role and permissions.

Zoe Subscription Intelligence agent is trained as a billing expert, with structured logic and billing context providing finance-grade results on billing and revenue recognition questions. It also has access to Zone's Knowledge Base and other foundational information, providing subject matter expertise across ZoneBilling.

Zone Control Center landing page with Zoe AI chat skill open in the corner.

How do I engage with Zoe Subscription Intelligence?

Each Zoe AI agent, like Subscription Intelligence, uses a plain language conversational chat to engage with our customers on the frontend, but secure and robust deterministic AI on the backend. This means the responses are reliable and less likely to hallucinate than generative AI like ChatGPT.

To use Zoe Subscription Intelligence, just ask questions in the 'chat'. Each submission counts as a single query - and you don't need to know SQL or how to build custom reports to get immediate answers.

Query Best Practices

Be specific and add parameters
Use date ranges
Verify responses
Iterate with follow-up questions
Reset when necessary
Split new conversations into new chat sessions

Don't	Do
Ask: "how many bills are pending?"	Ask: "how many bills are currently pending approval?"
Ask: "who paid late again?"	Ask: "who consistently pays late?"
Only Ask: "list the 10 customers who most frequently pay late"	Follow Up With: "what is the average number of days past due for each of the top 10 customers this year?" "what percentage of their total invoices are paid late?" "On average, how much revenue is tied up in accounts receivable from these top 10 customers?"

Response Elements

When using our Zoe's AI agents, there are three or four key elements in each query response: Summary Response, Results, Extended Analysis, and Suite QL Query.

Available Metrics & Functionality

The current scope of Zoe Subscription Intelligence is limited to operational metrics and will not include business metrics or data visualizations. This is because the use of business metrics is too variable across companies to provide a consistently accurate experience for our customers.

Zoe Subscription Intelligence uses specific fields sourced from Saved Searches on Charge records to provide real-time snapshots of a subscription's financial status including:

Amount Billed: The total sum of charges that have successfully been placed on NetSuite transactions.
Amount Pending Billing: The total of charges from past periods up through today that have not yet been invoiced.
Future Billing: The total of charges that have Bill Dates scheduled after the current date.
Amount Forecast: The overall forecasted value of upcoming charges.
Note that Bill Date is the key field used to determine when charges are scheduled.

Subscription Intelligence Uses Cases & Starter Prompts

Not sure where to get started? No problem! Watch this 3 minute video and check out the examples below.

Here are a few examples for different finance roles.

Scenario	Prompt Ideas for Zoe Subscription Intelligence
A revenue manager is asked to prepare for a board meeting.	Give me a breakdown of deferred revenue expected to recognize over the next 12 months, grouped by customer. Rank our top 10 customers by recognized revenue this quarter and show me how each one's revenue trend has moved over the past four quarters. Build me a deferred revenue rollforward for the quarter — showing opening balance, amounts billed, revenue recognized, and closing balance, written up in a format I can drop straight into the board deck. Reconcile our total billings to recognized revenue for this period and give me a clean summary of the difference, broken down into what's newly deferred, what's been released, and what's still outstanding.
A controller needs to complete the quarterly subscription performance comparison.	Analyze subscription performance for Q1 2026, broken down by billed and unbilled revenue. Compare Q1 2026 to the same Q1 2025 totals and provide insight. Break down the performance by subsidiary and customer, and analyze the results in more detail. Provide a strategic overview. What is the primary driver of growth in Q1 2026 and identify the single biggest risk we should monitor in the coming quarter.
A billing manager needs to evaluate customer engagement and financial risk.	Show me a list of customers who have consumed more than 80% of their prepaid balance for any items this quarter. Show me which customers have incurred the most overages in this quarter. Identify any customers whose current usage is trending 20% below their commit level. Who is at the highest risk for non-renewal. Provide a high-level overview of the information during this chat session, including summary statistics and insights.
A billing team member needs to complete a leakage analysis.	Show all unbilled subscriptions due this month. List subscriptions that have usage variable rate type but no corresponding usage detail. Has any rate type changed from form fixed to variable this year? Show me customers who consistently pay late and their credit terms.

Conversational Memory

Zoe AI agents were built with key memory elements to

leverage all information you provide within each conversation
only remember key information between conversations

This was done to maximize the reliability of responses while minimizing the need to repeat prompts.

Why does this matter?

You can ask Zoe AI agents to "remember" critical information and they will remember this next time you return the agent and start a new task. This saves time and prompting.

Within a conversation

Within a conversation (series of chats), Zoe agents will remember the information you've provided. For example, you can ask Zoe to provide you with a list of customers who consumed more than 80% of their prepaid balance for the quarter. Once you have that list, you can then ask her to identify the customers from that list who have consistently had overdue bills in the last 2 years. The same way you might have a conversation with a person who would remember what you've just told them.

Most of the contextual awareness will reset each time a new chat is opened - unless customers ask Zoe to remember critical information such as preferences, account specific details, or workflows.

A customer’s chat history does not need to be cleared to “reset” the contextual awareness for each new session or conversation.

Key information remembered between conversations

Zoe builds a personal notebook about each customer, including preferences, account specific details, and workflows. This means that you don't need to re-explain you setup each time you start a new chat. And answers are tailored to your preferences.

Zoe can remember key things such as your role and responsibilities, preferences for answer format and detail level, quirks about your setup, and the type of work you complete.

How can I help Zoe remember or forget information?

While Zoe will pick up on patterns and draw her own conclusions, you can help control what she remembers by asking her to remember critical information during a normal conversation. Her memory is yours to manage, so if your role changes, preferences shift, or responses aren't quite what you're expecting - ask "what do you remember about me"? Then correct anything that's off by asking to delete specific memories or clear everything and start fresh.

Related to

	Summary: Quick hitting response to the specific query. In the example below, there are 282 bills regularly paid late and the response lists the most frequent late-paying customers.
	Results: When responses include a list of at least 15 items, a separate “Results” element is provided to view the items in a table. When there are fewer than 15 items, the list can be viewed directly within the Extended Analysis. And in both cases, lists can be downloaded via CSV by selecting the download icon beneath all the response elements.
	Extended Analysis: Responses always include an Extended Analysis - a longer analysis that looks at patterns, segmentation, and includes key findings and recommendation for next steps.
	Suite QL Query: The last part of every response is what makes this “Explainable AI". The Suite QL Query shows exactly how each request is completed and provides an opportunity to verify that AI interpreted the query as intended.