Coauthored by: Venu Amancha (Lucid Data Hub) and Roy Hasson (Microsoft)
Lucid Data Hub is proud to announce Agent Mart Studio for Microsoft Fabric, enabling business users to build and deploy AI agents directly on top of enterprise data in OneLake. This integration enables organizations to automate complex business processes using proven AI agents, enhanced by industry-specific insights and safeguarded by contextual guardrails that ensure measurable business value.
In this post, we’ll outline the challenges many enterprises face when building AI agents, explore real-world use cases where AI agents are useful, walk through a simple agent-building workflow, and highlight how adaptive customization ensures long-term business impact.
Solving the Enterprise Data Bottleneck
For many organizations, getting actionable insights from enterprise data remains a challenge. Business users often depend on technical experts and complex tooling, causing delays, inefficiencies, and missed opportunities. AI agent creation has traditionally been restricted to engineering or data science teams, leaving non-technical teams without the ability to act directly on data.Lucid’s integration with Microsoft Fabric addresses this pain point. By combining the openness and interoperability of OneLake with Agent Mart Studio’s no-code environment. Business users can create AI agents using all of their enterprise data that automate workflows and surface insights in real time. Because every enterprise is different, these agents are designed to be both ready-to-use and configurable - ensuring they align with the specific processes and objectives of each organization.
https://www.youtube.com/watch?v=K8gVLS5-7VU
A Real-World Example: Retail Out of Stock
Consider a retail firm that needs to manage and respond to out-of-stock situations. Traditionally, this required manual tracking across sales reports, coordination between inventory managers and IT teams, and custom reporting to identify patterns. By the time insights are available, shelves are already empty and the chance to capture sales is gone.With Lucid’s Agent Mart Studio integrated into Microsoft Fabric, the same firm can empower its business analysts to automate out-of-stock detection using events stored and updated in real-time in OneLake. The agents can identify products at risk of going out of stock, generate detailed replenishment reports, and trigger real-time alerts. This allows supply chain teams to respond immediately—without writing a single line of code.
Importantly, the platform doesn’t assume a one-size-fits-all approach. Out-of-the-box, the agents deliver value and can be customized to match a retailer’s alert thresholds, monitored items, and replenishment rules. This balance of usability and adaptability ensures the solution is both practical today and sustainable for the future.
Hands-On Walkthrough: Automating Retail Out of Stock
Example: How a retail firm can build its first Product out-of-stock AI Agent using Lucid Data Hub and Microsoft Fabric.Step 1: Start with Adding Workload
From your Microsoft Fabric Workloads page, start by adding the Agent Mart Studio workload to your Fabric Workspace(s). This makes the AI Agent Builder app available right inside Fabric Workspace(s).
Step 2: Design your Agent
Launch Agent Mart Studio from the Microsoft Fabric Workload Hub. Use the intuitive drag-and-drop interface to define the agent’s business logic. This includes setting rules for managing out-of-stock retail scenarios—such as notifying relevant teams, triggering replenishment actions, or adjusting promotions based on inventory levels. Extend the logic further to suit your business needs, like sending automated order placement to vendors, building strategies for minimizing shrinkage and more.
Add a New 'AI Agent' Item: Select New Item, search for AI Agent under Agent Mart Studio section.
Enter the AI Agent Name and select Create.
Describe your AI Agent: Name your AI Agent and describe its purpose.
Example: ‘Create an AI Agent that monitors product availability, detects potential out-of-stock situations, and triggers replenishment recommendations to supplier'.
Data Tools Selection: Agent Mart scans your OneLake data and presents Data Tools—ready-to-use query blocks that answer specific business questions. Select the most relevant to your scenario, like inventory summary, Product Inventory vs Sales, Thresholds, etc.
Business Logic Selection: Next, Agent Mart Studio will suggest business logic flows. These are logic blocks designed to meet your requirement. Select & customize the logic to meet your requirements.
Customize Business Logic: You can customize logic—add SQL conditions, plug in a Python notebook for deeper analysis, or use LLM inference to generate narrative summaries.
Agent Action Selection: Finally, choose the actions your agent should take. Send an email alert, post to Teams, publish a Power BI dashboard, or call external services. Multiple actions can run in parallel.
Optional step: Scaling to support large number of datasets
Often, in larger organizations, it’s easier to create and model datasets in advance, allowing users to select AI ready data for their agents and reducing the steps needed to get started. Lucid Context Hub in Microsoft Fabric lets users select a Lakehouse in OneLake, such as one containing sales, product, or vendor data. It then automatically models and annotates the data for agent use. This step can also be done in-line as described in the previous step but completing it in advance reduces the steps users need to take to get started building AI agents.Step 3: Deploy and Test
Activate the agent and choose from built-in actions. These include creating marketing reports, updating dashboards with real-time insights, and sending out-of-stock reports to suppliers via email or Microsoft Teams. Additionally, the agent can connect to the order system to automatically trigger restocks, ensuring smooth operations. Test with sample sales & product inventory data and monitor the results through built-in dashboards in Agent Mart Studio’s Fabric application. Evaluate results in real-time and quickly iterate, refining replenishment criteria and workflows to enhance your retail store operations strategies.
Agent Publishing: When everything is ready, select 'Publish'. Your agent is now live.
Running the agent: You can run the agent instantly, monitor execution logs, and even schedule it to run automatically. Within seconds, insights and actions are delivered to your business teams.
With Agent Mart Studio in Microsoft Fabric, building intelligent AI agents is as simple as 1-2-3: Add → Design → Deploy. Your next AI agent is just a few clicks away with by unique industry knowledge and protected by contextual guardrails that ensure tangible business value.
Why It Matters
Agent Mart Studio’s integration with Microsoft Fabric and OneLake simplifies AI agent creation, enabling meaningful business impact with minimal engineering effort. By lowering the technical barriers, business users gain direct access to AI innovation using their enterprise data stored in OneLake. They can act faster, respond more strategically, and iterate quicklym, while minimizing reliance on long engineering cycles.At the same time, no two organizations are exactly alike. Agent Mart Studio has been designed to embrace that reality. Its intuitive setup delivers quick value, while the flexible design lets enterprises tailor agents to fit their workflows, logic, and tools. This combination of simplicity and adaptability ensures that enterprises can adopt AI responsibly and continue to derive value as their needs evolve.