1. What Insight Agent Is
Insight Agent is Xnurta's intelligent assistant for advertising analysis and strategic recommendations. It is not just a data query tool — it is an operational partner that helps users diagnose root causes and make high-quality decisions.
Core capabilities: broad and flexible intent recognition, reliable data retrieval, and fluid multi-turn conversation. Ask what you want to know, follow up naturally, and go as deep as you need. Current data coverage: Amazon Sponsored Ads.
📢 Insight Agent is currently available in Open Beta for eligible users. Users with access will see an Open Beta entry popup after logging in.
Beta popup | Main interface |
 |  |
2. Why It Is Different
Three things define the product experience.
2.1 Conversational, not template-bound
Users are not limited to a fixed workflow. They can start with a broad question, ask follow-up questions naturally, narrow the scope, and explore the answer step by step.
2.2 Trusted data grounding
Amazon advertising data is structurally complex. Similar business terms can refer to different entities depending on context, such as campaign spend, keyword spend, or ASIN spend. Insight Agent is designed to retrieve the right data context so answers are useful and trustworthy.
2.3 Built for diagnosis and decision support
The goal is not just to describe performance, but to help advertisers understand what changed, why it changed, and what to do next.
3. Core Use Cases
Insight Agent supports a range of question types, from simple lookups to deeper diagnosis and recommendation.
Use case | What it helps with |
Simple query | Check a metric, entity, or performance condition |
Overview analysis | Scan an account and surface what matters most |
Root-cause diagnosis | Understand why performance changed |
Strategy and recommendation | Identify next actions based on account data |
AI managed group performance review | Review AI managed group performance and explain recent AI actions |
Example questions
"Which campaigns drove the most orders last month?"
"Scan my account for problem ASINs with high ACOS or rising TACOS."
"My ACOS increased over the last two weeks. What changed?"
"Which campaigns should receive more budget based on recent performance?"
"How did my AI managed group perform last week, and what actions did the AI take?"
4. Current Scope
Insight Agent currently supports Amazon Sponsored Ads, including:
Sponsored Products
Sponsored Brands
Sponsored Display
It can analyze data across the main advertising entities, including:
Portfolios
Campaigns
Ad groups
Products, Parent ASINs, ASINs, and SKUs
Keywords and targets
Search terms
It also supports standard advertising metrics such as spend, sales, orders, impressions, clicks, ACOS, ROAS, CTR, CVR, CPC, and TACOS.
In addition to performance analysis, Insight Agent can help interpret account structure, settings, and selected optimization history relevant to campaign review.
5. Quality and Trust
We evaluate Insight Agent across two layers of quality:
Data accuracy — whether the underlying numbers are correct
Answer quality — whether the analysis is clear, well-reasoned, and actionable
Data accuracy comes first. If the numbers are wrong, the analysis is not useful.
In internal evaluations on real advertiser scenarios, Insight Agent outperformed general-purpose LLM + Amazon MCP workflows, including Claude, GPT, and Gemini setups, on data accuracy and overall output reliability. That matters because answers need to start from the right numbers before analysis quality can be trusted.
Like any AI product, Insight Agent should be used as decision support rather than an unreviewed replacement for human judgment. We continuously improve the system through evaluation, product refinement, and customer feedback.
6. AI Managed Group Performance Review and Explainability
Insight Agent can help review AI managed group performance and explain recent AI behavior in the context of account performance and optimization logic.
This is especially useful when customers want to understand what happened, why a change may have been made, and how recent optimization activity connects to campaign results. The current experience is strongest for conversational explanation and review, rather than a full low-level replay of every system event.
7. What Is In Development
The following capabilities are on the roadmap or in development:
Direct action from within the same conversational workflow
Broader data coverage, including AMC and other advertiser-relevant sources
Stronger long-term context across sessions
Richer reporting and file-based workflows
These areas are active priorities, but external delivery dates are not committed in this document.
8. Frequently Asked Questions
Q1. How is Insight Agent different from Reporting Agent?
Reporting Agent works best when the question fits a predefined template. Insight Agent is different: users can ask follow-up questions naturally, go deeper, and move from a surface question to diagnosis and recommendation in one conversation.
Q2. Why not just use a general LLM with Amazon MCP?
General-purpose LLM workflows can be useful for narrow technical use cases, but the trade-off is usually data reliability, stability, and usability. Insight Agent is purpose-built for advertiser questions and grounded retrieval, so the experience is more dependable for day-to-day analysis work.
Q3. Why is the accuracy not 100%?
No AI system handling open-ended natural-language analysis is perfect. Real advertiser questions are varied and often complex. Our approach is to measure quality continuously, improve against real usage, and treat trusted data as the foundation of the product.
Q4. Can Insight Agent take actions directly in the account?
Not today. The current product focuses on analysis, diagnosis, explanation, and recommendation. Direct action is in development.
Q5. Does Insight Agent support AMC?
Not natively today. AMC is part of our broader data coverage roadmap, and we are expanding carefully with a focus on quality and customer value.
Q6. Can Insight Agent review AI managed group performance and explain optimization decisions?
Yes. It can help review AI managed group performance and explain recent AI behavior in the context of account performance and optimization logic, which makes it useful for review, discussion, and decision support.
Q7. Does Insight Agent remember account context across sessions?
Longer-term context across sessions is in development. Today, the experience is strongest within the active conversation, with stronger persistent context planned over time.
9. Summary
Insight Agent helps advertisers move faster from question to understanding to action planning. It combines conversational ease of use with account-level analysis, practical recommendations, and a strong focus on trusted data.
For teams that want a faster way to understand performance, diagnose issues, and review optimization activity without building every answer manually, Insight Agent is designed to become a daily analysis companion.