In Microsoft Dynamics 365 CRM, Business Process Flows (BPFs) are powerful tools that guide users through defined business stages. However, when working with Lead-based BPFs that persist into Opportunity, certain platform limitations surface-especially when multiple Lead-rooted BPFs are involved. This blog walks through a real-world problem I encountered with Lead → Opportunity BPF switching, why the out-of-the-box behavior falls short, and how I designed a robust client-side + server-side solution to safely and reliably switch BPFs-even after the Lead has already been qualified. How BPFs Work (Quick Recap) It ideally won’t allow a switch, either via brute forcing via client side or server side as – The Problem In my scenario: The Challenge Once a Lead is qualified: This is non-intuitive, error-prone, and inefficient, especially considering the manual effort that goes into it. Solution Overview I implemented a guided, safe, and reversible BPF switching mechanism that: High-Level Architecture This solution uses: Step-by-Step Methodology 1. Entry Point: Opportunity Ribbon Button A custom ribbon button on the Opportunity form: These fields act as a controlled handshake between Opportunity and Lead. 2. Lead OnLoad: Controlled Trigger Execution On Lead form load: if (diffSeconds > 20) { return;} Xrm.WebApi.updateRecord(“lead”, formContext.data.entity.getId(), { cf_shouldtrigger: false}); This ensures: 3. Identifying and Aborting the Existing BPF Before switching: var activeProcess = formContext.data.process.getActiveProcess(); Xrm.WebApi.updateRecord(bpfEntityName,result.entities[0].businessprocessflowinstanceid,{statecode: 1, // Inactivestatuscode: 3 // Aborted}); This is a critical step—without aborting the old instance, Dynamics can behave unpredictably. 4. Switching the UI BPF After aborting: 5. Handling BPF Instance Creation (First-Time Switch Logic) The solution explicitly checks: If it exists: If it does NOT exist (first switch): This dual-path logic makes the solution idempotent and reusable. 6. Server-Side Plugin: Persisting the Truth A plugin ensures that: // Identify BPF typebool isNewBpf = (context.PrimaryEntityName == “new_bpf_entity”); // Resolve related LeadGuid leadId = isNewBpf? ((EntityReference)target[“bpf_leadid”]).Id: ((EntityReference)target[“leadid”]).Id; // Retrieve related Opportunity via LeadEntity opportunity = GetOpportunityByLead(service, leadId); // Determine stages and pathstring qualifyStageId = isNewBpf ? NEW_QUALIFY_STAGE : OLD_QUALIFY_STAGE;string finalStageId = isNewBpf ? NEW_FINAL_STAGE : OLD_FINAL_STAGE;string traversedPath =START_STAGE + “,” + qualifyStageId + “,” + finalStageId; // PATCH 1 – Qualify stageservice.Update(new Entity(target.LogicalName, target.Id){[“activestageid”] = new EntityReference(“processstage”, new Guid(qualifyStageId)),[“traversedpath”] = START_STAGE + “,” + qualifyStageId}); // PATCH 2 – Final stage + Opportunity bindservice.Update(new Entity(target.LogicalName, target.Id){[“activestageid”] = new EntityReference(“processstage”, new Guid(finalStageId)),[“traversedpath”] = traversedPath,[isNewBpf ? “bpf_opportunityid” : “opportunityid”] =new EntityReference(“opportunity”, opportunity.Id)}); // Mark Lead as successfully processedservice.Update(new Entity(“lead”, leadId){[“cf_pluginsuccess”] = new OptionSetValue(1) // Yes}); This guarantees data consistency and auditability. 7. Final UI Sync & Redirect After successful completion:   Xrm.Navigation.openForm({ entityName: “opportunity”, entityId: opportunityId }); From the user’s perspective: “I clicked a button, confirmed the switch, and landed back in my Opportunity—done.” Why This Solution Works ✔ Respects Dynamics 365 BPF constraints✔ Prevents orphaned or conflicting BPF instances✔ Handles first-time and repeat switches✔ Ensures server-side persistence✔ Minimal user disruption✔ Fully reversible Most importantly, it bridges the gap between platform limitations and real business needs. Final Thoughts Dynamics 365 BPFs are powerful—but when multiple Lead-rooted processes coexist, manual switching is not enough. This solution demonstrates how: can be combined to deliver a seamless, enterprise-grade experience without unsupported hacks. If you’re facing similar challenges with Lead → Opportunity BPF transitions, this pattern can be adapted and reused with confidence. We hope you found this blog useful, and if you would like to discuss anything, you can reach out to us at transform@cloudfronts.com

Power BI is excellent at visualizing insights, but insights often need action. That’s where the Power Automate visual comes in. With this visual, report consumers can trigger instant Power Automate flows directly from a Power BI report, using the data and filters already applied on the page. No switching tools. No exporting data. Just click and act. This blog walks through how the Power Automate visual works, how to configure it, and what to consider before rolling it out. Understanding Power Automate Visuals – The Power Automate visual adds a button to your Power BI report. When clicked, it runs an instant cloud flow. Key capabilities: From a user’s perspective, it feels like a native action button inside Power BI. Adding the Power Automate Visual In Power BI Desktop One can add the visual in two ways: Once added, the visual appears on the report page with built-in instructions. In Power BI Service The process is identical: One can resize or reposition the button like any other visual. Choosing the Flow Environment Before creating or attaching a flow, select the environment where the flow will live. The environment picker: Choosing the right environment upfront avoids permission and governance issues later. Making the Flow Data-Contextual One of the most powerful features of the Power Automate visual is data context. How it works Example: This makes flows responsive to how users are interacting with the report. Creating or Editing the Flow Editing from Power BI Desktop or Service With the flow selected, add any data fields to the Power Automate Data region, to use as dynamic inputs for the flow. Select More options (…) > Edit to configure the button. In edit mode of the visual, either select an existing flow to apply to the button, or create a new flow to apply to the button. One can start from scratch or start with one of the built-in templates as an example. To start from scratch, select New > Instant cloud flow. Select New step. Here, one can choose a subsequent action or specify a Control if you want to add more logic to determine the subsequent action. Optionally, one can reference the data fields as dynamic content if they want the flow to be data contextual. This example uses the Region data field to create an item in a SharePoint list. Based on the end-user’s selection, Region could have multiple values or just one. After you configure your flow logic, name the flow, and select Save. Select the arrow button to go to the Details page of the flow you created. Here’s the Details page for a saved flow. Select the Apply button  to attach the flow you created to your button. Formatting the Button The Power Automate button is fully customizable: This allows the button to match your report’s design and UX standards. Test the flow After the flow is applied to the button, we need to test it before you share the flow with others. These Power BI flows can only run in the context of a Power BI report. Thus one can’t run these flows in a Power Automate web app or elsewhere. If the flow is data contextual, make sure to test how the filter selections in the report affect the flow outcome. Sharing the Flow with Report Users When the flow runs successfully, it can be shared concerned personas of the report. Give users edit access Alternatively, you can give any users edit access to the flow, not just run permissions. Considerations and Limitations Before adopting the Power Automate visual, keep these points in mind: These constraints help maintain performance, security, and governance. When to Use the Power Automate Visual This pattern works best when you want to: In short, it bridges the gap between analysis and execution. Final Thoughts The Power Automate visual transforms Power BI from a read-only analytics tool into an interactive action surface: Analyze → Filter → Click → Automate When used thoughtfully, it empowers users to act on insights at the exact moment they discover them — without breaking their flow. We hope you found this blog useful, and if you would like to discuss anything, you can reach out to us at transform@cloudfronts.com

Managing and processing large datasets efficiently is a key requirement in modern data engineering. Azure Databricks, an optimized Apache Spark-based analytics platform, provides a seamless way to handle such workflows. This blog will explore how PySpark and SQL can be combined to dynamically process, and clean data using the medallion architecture (Only Raw → Silver) and store the results in Azure Blob Storage as PDFs. Understanding the Medallion Architecture: – The medallion architecture follows a structured approach to data transformation: Aggregated Layer (Gold): Optimized for analytics, reports, and machine learning. In our use case, we extract raw tables from Databricks, clean them dynamically, and store the refined data into the silver schema. Key technologies / dependencies used: – Step-by-Step Code Breakdown 1. Setting Up the Environment Install & import necessary libraries The above command installs reportlab, which is used to generate PDFs. This imports essential libraries for data handling, visualization, and storage. 2. Connecting to Azure Blob Storage This snippet authenticates the Databricks notebook with Azure Blob Storage and prepares a connection to upload the final PDFs; Initiates the Spark Session as well. 3. Cleaning Data: Raw to Silver Layer Fetch all raw tables This dynamically removes NULL values from raw data and creates a cleaned table in the silver layer. 4. Verifying and comparing the Raw and the Cleaned (Silver) 4. Converting Cleaned Data to PDFs 5. Converting Cleaned Data to PDFs Output at the Azure Storage Container This process reads cleaned tables, converts them into PDFs with structured formatting, and uploads them to Azure Blob Storage. 6. Automating cleaning at Databricks at fixed scheduleThis is automated by scheduling the notebook & it’s associated compute instance to run at fixed intervals and timestamps. Further actions: – Why Store Data in Azure Blob Storage? To conclude, by leveraging Databricks, PySpark, SQL, ReportLab, and Azure Blob Storage, we have automated the pipeline from raw data ingestion to cleaned and formatted PDF reports. This approach ensures: a. Efficient data cleansing using SQL queries dynamically. b. Structured data transformation within the medallion architecture. c. Seamless storage and accessibility through Azure Blob Storage. This methodology can be extended to include Gold Layer processing for advanced analytics and reporting. We hope you found this blog useful, and if you would like to discuss anything, you can reach out to us at transform@cloudFronts.com

In Microsoft Dynamics 365, subgrids are a powerful way to display related records on a form. But what happens when: Out-of-the-box, Dynamics 365 doesn’t give us many options here. We can select a view, but we cannot apply dynamic filters unless the entities are directly related or the criteria already exist in the view’s FetchXML. This is where the JavaScript setFilterXml() API becomes a life-saver. In this article, I’ll show you how to filter a subgrid dynamically using JavaScript — even when the subgrid’s entity is completely unrelated to the main form entity. Use Case Imagine your form has a field called Name, and you want to filter the subgrid so that it shows only records whose Name begins with the same prefix. But: As there are also records, where the lookup column might need to be empty on purpose, which further would break relationship based filtering in the subgrid. OOB? Impossible. With JavaScript? Totally doable. How the JS based subgrid filtering works In Dynamics 365, subgrids are rendered as independent UI components inside the form. Even though the form loads first, subgrids load asynchronously in the background, which means: The form and its fields may already be available, but the subgrid control might not yet exist, so trying to apply a filter immediately on form load will fail. Here is the basic structure of a JS Function to perform Subgrid filtering – This control represents the interactive UI component that displays the records for the view.It gives you programmatic access to:-> Set filters-> Refresh the grid-> Access its view ID-> Handle events (in some versions) However, because subgrids load later than the form, this line may return null the first several times. If you proceed at that point, your script will break.So we implement a retry pattern: If the subgrid is not ready, wait 100ms -> Try again -> Repeat until the control becomes availableThis guarantees that our next steps run only when the subgrid is fully loaded. var oAnnualTCVTargetGridFilter = oAnnualTCVTargetGridFilter || {}; oAnnualTCVTargetGridFilter.filterSubgrid = function(executionContext) {var formContext = executionContext.getFormContext(); }; To make sure the filter is applied correctly, we follow a three-step workflow: 1. Retry Until the Subgrid Control Is Loaded (setTimeout) – When the script runs, we attempt to retrieve the subgrid control using: var subgrid = formContext.getControl(“tcvtargets”); 2. Apply the Filter (setFilterXml()) – Once the subgrid control is found, we can safely apply a filter. Then we can apply our filtering logic, and utilize it in the FetchXML Query: -> Read the field Name (cf_name) from the main form & design a logic -> Construct a FetchXML <filter> element -> Passing this XML to the subgrid using: This tells Dynamics 365 to apply an additional filter on top of the existing view used by the subgrid. A few important things to note: If the cf_name field is empty, we instead apply a special filter that returns no rows. This ensures the grid displays only relevant and context-driven data. 3. Refresh the Subgrid (subgrid.refresh()) – After applying the filter XML, the subgrid must be refreshed: Without this call, Dynamics will not re-run the query, meaning your filter won’t take effect until the user manually refreshes the subgrid. Refreshing forces the system to: -> Re-query data using the combined view FetchXML + your custom filter -> Re-render the grid -> Display the filtered results immediately This gives the user a seamless, dynamic experience where the subgrid shows exactly the records that match the context. JS + FetchXML based filtering in action – Without filtering :- With filtering :- Key Advantages of This Approach Works Even When No Relationship Exists Possibility to filter a subgrid even if the target entity has no direct link to the form’s main entity. This is extremely useful in cases where the relationship must remain optional or intentionally unpopulated. Enables Dynamic, Contextual Filtering We can design filtering logic on the form field values, user selections, or business rules. Filtering on Fields Not Included in the View Since the filtering logic is applied client-side, there is no need to modify or clone the system view just to include filterable fields. Bypasses Limitations of Lookup-Based Relation Filtering This method works even when the lookup column is intentionally left empty, which is a scenario where OOB relationship-based filtering fails. More Flexible Than Traditional View Editing You can apply advanced logic such as prefix matching, conditional filters, or dynamic ranges—things not possible using standard UI-only configuration. To conclude, filtering subgrids dynamically in Dynamics 365 is not something the platform supports out-of-the-box- especially when the entities are unrelated or when the filter criteria doesn’t exist in the subgrid’s original view. However, with a small amount of JavaScript and the setFilterXml() API, you gain complete control over what data appears inside a sub grid, purely based on the context passed from the main form. I Hope you found this blog useful, and if you would like to discuss anything, you can reach out to us at transform@cloudFronts.com.

Working with Microsoft Dataverse Connector in Power BI is usually straightforward—until you encounter a table that simply refuses to load any rows, even though the data clearly exists in the environment. This happens especially with hidden, virtual, or system-driven tables (e.g. msdyn_businessclosure, msdyn_scheduleboardsetting) which are commonly used in Field Service and Scheduling scenarios. Before jumping to a workaround, it’s important to understand why certain Dataverse tables don’t load in Power BI, what causes this behavior, and why the standard Dataverse connector may legitimately return zero rows. Causes – 1] The Table Is a Virtual or System Table with Restricted AccessSystem-managed Dataverse tables like msdyn_businessclosure are not exposed to the Dataverse connector because they support internal scheduling and platform functions. 2] No Records Exist in the Root Business Unit Data owned by child business units is not visible to Power BI accounts associated with a different BU, resulting in zero rows returned. 3] The Table Is Not Included in the Standard Dataverse Connector Some solution-driven or non-standard tables are omitted from the Dataverse connector’s supported list, so Power BI cannot load them. Solution: Export Dataverse Data Using Power Automate + Excel Sync Since Power BI can read:-> OneDrive-hosted files-> Excel files-> SharePoint-hosted spreadsheets …a suitable workaround is to extract the restricted Dataverse table into Excel using a scheduled (When the records are few) / Dataverse triggered (When there are many records and you only want a single one, to avoid pagination) Power Automate flow. What it can do –-> Power Automate can access system-driven tables.-> Excel files in SharePoint can be refreshed by Power BI Service.-> we can bypass connector restrictions entirely.-> The method works even if entities have hidden metadata or internal platform logic. This ensures:-> Consistent refresh cycles-> Full visibility of all table rows-> No dependency on Dataverse connector limitations Use case I needed to use the Business Closures table (Dataverse entity: msdyn_businessclosure) for a few calculations and visuals in a Power BI report. However, when I imported it through the Dataverse Connector, the table consistently showed zero records, even though the data was clearly present inside Dynamics 365. There are 2 reasons possible for this –1] It is a System/Platform Tablemsdyn_businessclosure is a system-managed scheduling table, and system tables are often hidden from external connectors, causing Power BI to return no data. 2] The Table Is Not Included in “Standard Tables” Exposed to Power BIMany internal Field Service and scheduling entities are excluded from the Dataverse connector’s metadata, so Power BI cannot retrieve their rows even if they exist. So here, we would fetch the records via “Listing” in Power automate and write to an excel file to bypass the limitations that hinder the exposure of that data; without compromising on user privileges, or system roles; we can also control or filter the rows being referred directly at source before reaching PBI Report. Automation steps – 1] Select a suitable trigger to fetch the rows of that entity (Recurring or Dataverse, whichever is suitable). 2] List the rows from the entity (Sort/Filter/Select/Expand as necessary). 3] Perform a specific logic (e.g. clearing the existing rows, etc.) on the excel file where the data would be written to. 4] For each row in the Dataverse entity, select a primary key (e.g. the GUID), provide the path to the particular excel file (e.g. SharePoint -> Location -> Document Library -> File Name -> Sheet or Table in the Excel File), & assign the dynamic values of each row to the columns in the excel file. 5] Once this is done, import it to the PBI Report by using suitable Power Query Logic in the Advanced Editor as follows – -> a) Loading an Excel File from SharePoint Using Web.Contents() – Source = Excel.Workbook(Web.Contents(“https://<domain>.sharepoint.com/sites/<Location>/Business%20Closures/msdyn_businessclosures.xlsx”),null,true), What this step does: -> Uses Web.Contents() to access an Excel file stored in SharePoint Online.-> The URL points directly to the Excel file msdyn_businessclosures.xlsx inside the SharePoint site.-> Excel.Workbook() then reads the file and returns a structured object containing:All sheets, Tables, Named ranges Parameters used: null → No custom options (e.g., column detection rules)true → Indicates the file has headers (first row contains column names) -> b) Extracting a Table Named “Table1” from the Workbook – msdyn_businessclosures_Sheet = Source{[Item=”Table1″, Kind=”Table”]}[Data], This would search inside the Source object (which includes all workbook elements), and look specifically for an element where: Item = “Table1” → the name of the table in the Excel fileKind = “Table” → ensures it selects a table, not a sheet with the same name & would extract only the Data portion of that table. As a result, we get Power Query table containing the exact contents of Table1 inside the Excel workbook, to which we can further apply our logic filter, clean, etc. To conclude, when Dataverse tables refuse to load through the Power BI Dataverse Connector—especially system-driven entities like msdyn_businessclosure—the issue is usually rooted in platform-level restrictions, connector limitations, or hidden metadata. Instead of modifying these constraints, offloading the data through Power Automate → Excel → Power BI provides a controlled, reliable, and connector-independent integration path. By automating the extraction of Dataverse rows into an Excel file stored in SharePoint or OneDrive, you ensure: This method is simple to build, stable to maintain, and flexible enough to adapt to any Dataverse table -whether standard, custom, or system-managed. For scenarios where Power BI needs insights from hidden or restricted Dataverse tables, this approach remains one of the most practical and dependable solutions. I Hope you found this blog useful, and if you would like to discuss anything, you can reach out to us at transform@cloudFronts.com.

Why SmartPitch? – The Idea and Pain Point  The idea for SmartPitch came directly from observing the day-to-day struggles of sales and pre-sales teams. Every Marketing Qualified Lead (MQL) to Sales Qualified Lead (SQL) conversion required hours of manual work: searching through documents stored in SharePoint, combing through case studies, aligning them with solution areas, and finally packaging them into a client-ready pitch deck.  The reality was that documents across systems—SharePoint, Dynamics 365, PDFs, PPTs—remained underutilized because there was no intelligent way to bring them together.   Sales teams often relied on tribal knowledge or reused existing decks with limited personalization.  We asked: What if a sales assistant could automatically pull the right case studies, map them to solution areas, and draft an elevator pitch on demand, in minutes?  That became the SmartPitch vision: an AI-powered agent that:  As a result of this product, it has helped us reduce pitch creation time by 70%.   2. The First Prototype – Custom Copilot Studio  Our first step was to build SmartPitch using Custom Copilot Studio. It gave us a low-code way to experiment with conversational flows, integrate with Azure AI Search, and provide sales teams with a chat interface.  1. Knowledge Sources Integration  2. Data Flow  3. Conversational Flow Design  4. Integration and Security  5. Technical Stack  6. Business Process Enablement  7. Early Prototypes  With Custom Copilot, we were able to:  We successfully demoed these early prototypes in Zurich and New York. They showed that the idea worked but they also revealed serious limitations.  3. Challenges in Custom Copilot  Despite proving the concept, Custom Copilot Studio had critical shortcomings:  Lacked support for model fine-tuning or advanced RAG customization.  However, incorporating complex external APIs or custom workflows was difficult.  This limitation meant SmartPitch, in its Copilot form, couldn’t scale to meet enterprise standards.  4. Rebuilding in Azure AI Foundry – Smarter, Extensible, Connected  The next phase was Azure AI Foundry, Microsoft’s enterprise AI development platform. Unlike Copilot Studio, AI Foundry gave us:  Extending SmartPitch with Logic Apps  One of the biggest upgrades was the ability to integrate Azure Logic Apps as external tools for the agent. This allowed SmartPitch to:  This modular approach meant we could add new functionality simply by publishing a new Logic App. No redeployment of SmartPitch was required.  Automating Document Vectorization  We also solved one of the biggest bottlenecks—document ingestion and retrieval—by building a pipeline for automatic document vectorization from SharePoint:  This allowed SmartPitch to search across text, images, tables, and PDFs, providing relevant answers instead of keyword matches.  But There Were Limitations  Even with these improvements, we hit roadblocks:  At this point, we realized the true bottleneck wasn’t the agent itself, it was the quality of the data powering it.  5. Bad Data, Governance, and the Medallion Architecture  SmartPitch’s performance was only as good as the data it retrieved from. And much of the enterprise data was dirty: duplicate case studies, outdated documents, inconsistent file formats.  This led to irrelevant or misleading responses in pitches.  To address this, we turned to Databricks’ Unity Catalog and Medallion Architecture:  You can read our post on building a clean data foundation with Medallion Architecture [Link]   Now, every result SmartPitch surfaced could be trusted, audited, and tied to a governed source.  6. SmartPitch in Mosaic AI – The Final Evolution  The last stage was migrating SmartPitch into Databricks Mosaic AI, part of the Lakehouse AI platform. This was where SmartPitch matured into an enterprise-grade solution.  What We Gained in Mosaic AI:  In Mosaic AI, SmartPitch wasn’t just a chatbot it became a data-native enterprise sales assistant:  From these, we came to know the following differences between agent development in AI Foundry & DataBricks Mosaic AI –    Attribute / Aspect  Azure AI Foundry  Mosaic AI  Focus  Developer and Data Scientist  Data Engineers, Analysts, and Data Scientists  Core Use Case  Create and manage your own AI agent  Build, experiment, and deploy data-driven AI models with analytics + AI workflows  Interface  Code-first (SDKs, REST APIs, Notebooks)  No-code/low-code UI + Notebooks + APIs  Data Access  Azure Blob, Data Lake, vector DBs  Native integration with Databricks Lakehouse, Delta Lake, Unity Catalog, vector DBs  MCP Server  Only custom MCP servers supported; built-in option complex  Native MCP support with Databricks ecosystem; simpler setup  Models  90 models available  Access to open-source + foundation models (MPT, Llama, Mixtral, etc.) + partner models  Model Customization  Full model fine-tuning, prompt engineering, RAG  Fine-tuning, instruction tuning, RAG, model orchestration  Publish to Channels  Complex (Azure Bot SDK + Bot Framework + App Service)  Direct integration with Databricks workflows, APIs, dashboards, and third-party apps  Agent Update  Real-time updates in Microsoft Teams  Updates deployed via Databricks workflows; versioning and rollback supported  Key Capabilities  Prompt flow orchestration, RAG, model choice, vector search, CICD pipelines, Azure ML & responsible AI integration  Data + AI unification (native to Lakehouse), RAG with Lakehouse data, multi-model orchestration, fine-tuning, end-to-end ML pipelines, secure governance via Unity Catalog, real-time deployment  Key Components  Workspace & agent orchestration, 90+ models, OpenAI pay-as-you-go or self-hosted, security via Azure identity  Mosaic AI Agent Framework, Model Serving, Fine-Tuning, Vector Search, RAG Studio, Evaluation & Monitoring, Unity Catalog Integration  Cost / License  Vector DB: external, Model Serving: token-based pricing (GPT-3.5, GPT-4), Fine-tuning: case-by-case, Total agent cost variable (~$5k–$7k+/month)  Vector Search: $605–$760/month for 5M vectors, Model Serving: $90–$120 per million tokens, Fine-Tuning Llama 3.3: $146–$7,150, Managed Compute built into DBU usage, End-to-end AI Agent ~$5k–$7k+/month  Use Cases / Capabilities  Agents intelligent, can interact/modify responses; single AI search per agent; infrastructure setup required; custom MCP server registration  Agents intelligent, interact/modify responses; AI search via APIs (Google/Bing); in-built MCP server; complex infrastructure; slower responses as results batch sent together  Development Approach  Low-code, faster agent creation, SDK-based, easier experimentation  Manual coding using MLflow library, more customization, API integration, higher chance of errors, slower build  Models Comparison  90 models, Azure OpenAI (GPT-3.5, GPT-4), multi-modal  ~10 base models, OSS & partner models (Llama, Claude, Gemma), many models don’t support tool usage  Knowledge Source  One knowledge source of each type (adding new replaces previous)  No limitation; supports data cleaning via Medallion Architecture; SQL-only access inside agent; Spark/PySQL not supported in agent  Memory / Context Window  8K–128K tokens (up to 1M for GPT-4.1)  Moderate, not specified  Modalities  Text, code, vision, audio (some models)  Likely text-only  Special Enhancements  Turbo efficiency, reasoning, tool calling, multimodal  Varies per model (Llama, Claude, Gemma architectures)  Availability  Deployed via Azure AI Foundry  Through Databricks platform  Limitations  Only one knowledge source of each type, infrastructure complexity for MCP server  No multi-modal Spark/PySQL access, slower batch responses, limited model count, high manual development  7. Lessons Learned:  … Continue reading Inside SmartPitch: How CloudFronts Built an Enterprise-Grade AI Sales Agent Using Microsoft and Databricks Technologies  →

AI-powered agents are quickly becoming the round the clock assistants of modern enterprises. They automate workflows, respond to queries, and integrate with data sources to deliver intelligent outcomes. But what happens when your agent needs to extend its abilities beyond what’s built-in? That’s where Logic Apps come in. In this blog, we’ll explore how you can add functionality to an AI Foundry Agent by connecting it with Azure Logic Apps-turning your agent into a truly extensible automation powerhouse. Why Extend an AI Foundry Agent? AI Foundry provides a framework to build, manage, and deploy AI agents in enterprise environments. By default, these agents can handle natural language queries and interact with pre-integrated data sources. However, business use cases often demand more: To achieve this, you need a bridge between your agent and external systems. Azure Logic Apps is that bridge. Enter Logic Apps Azure Logic Apps is a cloud-based integration service that enables you to: When integrated with AI Foundry Agents, Logic Apps can serve as external tools the agent can call dynamically. Steps to achieve external integrations / extending functionality in AI Foundry Agents with Logic Apps :- 1] Assuming your Agent Instructions and Knowledge Sources are ready, go to your Actions under Knowledge – 2] In the pop-up window, select Azure Logic Apps, you can also use other actions based on your requirement – 3] Here you would see a list of Microsoft Authored as well as our custom-built Logic App based Tools. To be displayed here, for suitable use by the AI Foundry Agent, it should meet a certain criterion as follows – a] Should be preferably on Consumption Plan, b] Should have an HTTP Request Trigger, atleast one Action, and a Response, c] In the Methods, select “Default (Allow all Methods)”, d] And a suitable description in the trigger, e] A Request Body (Auto Generated if created directly from AI Foundry). The developer can either create a Trigger from AI Foundry or, manually create a Logic App in the same Azure Subscription as the AI Foundry Project, observing the criteria. 4] As you can see below, For the scope of the blog I am covering a simple requirement of getting the list of clients for the SmartPitch Project, to fetch the case studies based on it; As you can see, the Logic App Tool meets the requirements for compatibility with Azure AI Foundry, with the required logic between the request and response. 5] As you can see below, For the scope of the blog I am covering a simple requirement of getting the list of clients for the SmartPitch Project, to fetch the case studies based on those;Once the Logic App is successfully created it would be visible in the Logic App Actions; select that Logic App to enable it as Tool. 6] Verify the details of the Logic App Tool and proceed. 7] Next you need to provide / verify the following information –a) Tool Name – The Name by which the Logic App would be accessible as a tool in the Agent, b) Connection to the Agent (Automatically assigned), c) Description to invoke the Tool (Logic App) – This is a crucial part for providing intent to the agent to when and how to use this logic app, and also what to expect from it. “Provide as much details as possible about the circumstances in which the tool should be called by the agent” 8] Once the Tool is created, it would be visible in the Actions list, and be ready for use. Here to check if the Intent is being understood and the tool being called, I have specifically instructed it to mention the name of the tool as well, along with it’s result. As you can see in the screenshot, the tool is triggered successfully, and the expected output is displayed. Example Use Case: Smart Pitch Agent Imagine your sales team uses an AI Foundry Agent (like “Smart Pitch Agent”) to create tailored pitches. By connecting Logic Apps, you can enable the agent to: Which we already have achieved in the in the above AI Agent using the other Logic App Tools The aim is to expose each capability as a Logic App, and the agent calls them as tools in conversation flow. Benefits of This Approach To conclude, by combining AI Foundry Agents with Azure Logic Apps, you unlock a powerful pattern: Together, they create a flexible, extensible solution that evolves with your enterprise needs. I Hope you found this blog useful, and if you would like to discuss anything, you can reach out to us at transform@cloudfronts.com.

The fastest way to lose control of enterprise data? Managing governance separately across workspaces. Unity Catalog solves this with one centralized layer for security, lineage, and discovery. Data governance is crucial for any organization looking to manage and secure its data assets effectively. Databricks’ Unity Catalog is a centralized solution that provides a unified interface for managing access control, auditing, data lineage, and discovery. This blog will guide you through the process of setting up Unity Catalog in your Databricks workspace. What is Unity Catalog? Unity Catalog is Databricks’ answer to centralized data governance. It enables organizations to enforce standards-compliant security policies, apply fine-grained access controls, and visualize data lineage across multiple workspaces. It ensures compliance and promotes efficient data management. Key Features: 1] Standards-Compliant Security: ANSI SQL-based access policies that apply across all workspaces in a region. 2] Fine-Grained Access Control: Support for row- and column-level permissions. 3] Audit Logging: Tracks who accessed what data and when. 4] Data Lineage: Provides visualization of data flow and dependencies. Unity Catalog Object Hierarchy Before diving into the setup, it’s important to understand the hierarchical structure of Unity Catalog: 1] Catalogs: The top-level container (e.g., Production, Development) that represents an organizational unit or environment. 2] Schemas: Logical groupings of tables, views, and AI models within a catalog. 3] Tables and Views: These include managed tables fully governed by Unity Catalog and external tables referencing existing cloud storage. Here is the procedure to setup a Unity Catalog Metastore in association with Azure Storage, as I have done for one of our products (SmartPitch Sales & Marketing Agent) – 1] First create a storage account  with primary service being – “Azure Blob Storage or Azure Data Lake Storage Gen 2”; Performance and Redundancy can be chosen based on the requirement for which the DataBricks service is being used.Here for my Mosaic AI Agent, I have used Locally Redundant Storage & Data Lake Gen 2 2] Once the storage account is created, ensure that you have enabled “Hierarchical Namespace” When creating a Unity Catalog metastore with Azure Blob Storage, Hierarchical Namespace (HNS) is required because Unity Catalog needs: a] Folder-like structure to organize catalogs, schemas, and tables. b] Atomic operations (rename, move, delete) on directories and files. c] POSIX-style access controls for fine-grained permissions. d] Faster metadata handling for lineage and governance. HNS turns Azure Blob into ADLS Gen2, which supports these features. 3] Upload any Raw/Unclean files to your metastore folder in the blob storage, which would be required for your use in DataBricks. 4] Create a Unity Catalog Connector in Azure Portal and assign it “Storage Blob Data Contributor” Role . 5] Assign CORS (Cross-Origin Resource Sharing) settings for that storage account. Why this is necessary: In short: Without configuring CORS, Databricks cannot communicate with your storage container to read/write managed tables, schema metadata, or logs. 6] Generate SAS Token   7] Navigate to your Workspace and select “Manage Account” – this should be done from the account admin.  8] Select Catalog tab on the left and then click “Create Metastore”    9] Assign a Name, Region (Same as Workspace), The path to the storage account, and the connector id. 10] Once the Metastore is created, assign it to a workspace .  11] Once this is done, the catalogs and the schemas, and tables in within it can be created. How does Unity Catalog differ from Hive Metastore ? Feature Hive Metastore Unity Catalog Scope Workspace or cluster-specific Centralized, spans multiple workspaces and regions Architecture Single metastore tied to Spark/Hive Cloud-native service integrated with Databricks Object Hierarchy Databases → Tables → Partitions Catalogs → Schemas → Tables/Views/Models Data Assets Supported Tables, views Tables, views, files, ML models, dashboards Security Basic GRANT/DENY at database/table level Fine-grained, ANSI SQL–based (catalog, schema, table, column, row) Lineage Not available Built-in lineage and impact analysis Auditing Limited or external Integrated audit logs across workspaces Storage Management Points to storage locations; no governance Manages external and managed tables with governance Cloud Integration Primarily on cluster storage or external path Secure integration with ADLS Gen2, S3, GCS Permissions Model Spark SQL statements Attribute- and role-based access, unified policies Use Cases Basic metadata store for Spark/Hive workloads Enterprise-wide data governance, sharing, and compliance To conclude, Unity Catalog is the next-generation governance and metadata solution for Databricks, designed to give organizations a single, secure, and scalable way to manage data and AI assets. Unlike the older Hive Metastore, it centralizes control across multiple workspaces, supports fine-grained access policies, delivers built-in lineage and auditing, and integrates seamlessly with cloud storage like Azure Data Lake, S3, or GCS. When setting it up, key steps include: 1] Creating a metastore and linking it to your workspaces. 2] Enabling hierarchical namespace on Azure storage for folder-level security and operations. 3] Configuring CORS to allow Databricks domains to interact with storage. 4] Defining catalogs, schemas, and tables for structured governance. By implementing Unity Catalog, you ensure stronger security, better compliance, and faster data discovery, making your Databricks environment enterprise-ready for analytics and AI. Business Outcomes of Unity Catalog By implementing Unity Catalog, organizations can achieve: Why now? As data volumes and regulatory requirements grow, organizations can no longer rely on fragmented or legacy governance tools. Unity Catalog offers a future-proof foundation for unified data management and AI governance—essential for any modern data-driven enterprise. At CloudFronts, we help enterprises implement and optimize Unity Catalog within Databricks to ensure secure, compliant, and scalable data governance for enterprise data governance.Book a consultation with our experts to explore how Unity Catalog can simplify compliance and boost productivity for your teams.Contact us today at Transform@cloudfronts.com to get started. To learn more about functionalities of DataBricks and other Azure AI services, please refer to my other blogs from the links given below: – 1] The Hidden Cost of Bad Data:How Strong Data Management Unlocks Scalable, Accurate AI – CloudFronts 2] Automating Document Vectorization from SharePoint Using Azure Logic Apps and Azure AI Search – CloudFronts 3] Using Open AI and Logic Apps to develop a Copilot agent for … Continue reading Setting Up Unity Catalog in Databricks for Centralized Data Governance →

In modern enterprises, documents stored across platforms like SharePoint often remain underutilized due to the lack of intelligent search capabilities. What if your organization could automatically extract meaning from those documents—turning them into searchable vectors for advanced retrieval systems? That’s exactly what we’ve achieved by integrating Azure Logic Apps with Azure AI Search. Workflow Overview Whenever a user uploads a file to a designated SharePoint folder, a scheduled Azure Logic App is triggered to: Once stored, a scheduled Azure Cognitive Search Indexer kicks in. This indexer: Technologies / resources used: –-> SharePoint: A common document repository for enterprise users, ideal for collaborative uploads. -> Azure Logic Apps: Provides low-code automation to monitor SharePoint for changes and sync files to Blob Storage. It ensures a reliable, scheduled trigger mechanism with minimal overhead. -> Blob Storage: Serves as the staging ground where documents are centrally stored for indexing—cheaper and more scalable than relying solely on SharePoint connectors. -> Azure AI Search (Cognitive Search): The intelligence layer that runs a skillset pipeline to extract, transform, and vectorize the content, enabling semantic search, multimodal RAG (Retrieval Augmented Generation), and other AI-enhanced scenarios. Why Not Vectorize Directly from SharePoint? Reference:-1. https://learn.microsoft.com/en-us/azure/search/search-howto-index-sharepoint-online2. https://learn.microsoft.com/en-us/azure/search/search-howto-indexing-azure-blob-storage How to achieve this? Stage 1: – Logic App to sync Sharepoint files to blob Firstly, create a designated Sharepoint directory to upload the required documents for vectorization. Then create the logic app to replicate the files along with it’s format and properties to the associated blob storage – 1] Assign the site address and the directory name where the documents are uploaded in Sharepoint – In the trigger action “When an item is created or modified”. 2] Assign a recurrence frequency, start time and time zone to check/verify for new documents and keep the blob container updated. 3] Add an action component – “Get file content using path”; and dynamically provide the full path (includes file extension), from the trigger 4] Finally, add an action to create blobs in the designated container that would be vectorized – provide the storage acc. name, directory path, the name of blob (Select to dynamically get the file name with extension for the trigger), blob content (from the get file content action). 5] On successfully saving & running this logic app, either manually or on trigger, the files are replicated in it’s exact form to the blob storage. Stage 2 :- Azure AI Search resource to vectorize the files in blob storage In Azure Portal (Home – Microsoft Azure), search for Azure AI Search service, and provide the necessary details, based on your requirement select a pricing tier. Once resource is successfully created, select “Import & vectorize data” From the 2 options – RAG and Multimodal RAG Index, select the latter one.RAG combines a retriever (to fetch relevant documents) with a generative language model (to generate answers) using text-only data. Multimodal RAG extends the RAG architecture to include multiple data types such as text, images, tables, PDFs, diagrams, audio, or video. Workflow: Now follow the steps and provide the necessary details for the index creation Enable deletion tracking, to remove the records of deleted documents from the index Provide a document intelligence resource to enable OCR, and to get location metadata for multiple document types. Select image verbalization (to verbalize text in images) or multimodal embedding to vectorize the whole image. Assign the LLM model for generating the embeddings for the text/images provide an image output location, to store images extracted from the files Assign a schedule to refresh the indexer and to keep the search index up to date with new documents. Once successfully created, search keywords in the search explorer of the index, to verify the vectorization, the results are provided based on it’s relevance and score/distance, to the user’s search query. Let us test this index in Custom Copilot Agent , by importing this index as an azure ai search knowledge source. On fetching details of certain document specific information, the index is searched for the most appropriate information, and the result is rendered in readable format by generative AI.  We hope you found this blog useful, and if you would like to discuss anything, you can reach out to us at transform@cloudfronts.com.

. Key Takeaways 1. Bad data kills ML performance — duplicate, inconsistent, missing, or outdated data can break production models even if training accuracy is high.2. Databricks Medallion Architecture (Raw → Silver → Gold) is essential for turning raw chaos into structured, trustworthy datasets ready for ML & BI.3. Raw layer: capture all data as-is; Silver layer: clean, normalize, and standardize; Gold layer: curate, enrich, and prepare for modeling or reporting.4. Structured pipelines reduce compute cost, improve model reliability, and enable proactive monitoring & feature freshness.5. Data quality is as important as algorithms — invest in transformation and governance for scalable, accurate AI. While developing a Machine Learning Agent for Smart Pitch (Sales/Presales/Marketing Assistant chatbot) within the Databricks ecosystem, I’ve seen firsthand how unclean, inconsistent, and incomplete data can cripple even the most promising machine learning initiatives. While much attention is given to models and algorithms, what often goes unnoticed is the silent productivity killer lurking in your pipelines: bad data. In the chase to build intelligent applications, developers often overlook the foundational layer—data quality. Here’s the truth: It is difficult to scale machine learning on top of chaos. That’s where the Raw → Silver → Gold data transformation framework in Databricks becomes not just useful, but also essential. The Hidden Cost of Bad Data Imagine you’ve built a high-performance ML model. It’s accurate during training, but underperforms in production. Why?Here’s what I frequently detect when I scan input pipelines:– Duplicate records– Inconsistent data types– Missing values– Outdated information– Schema drift (Schema drift introduces malformed, inconsistent, or incomplete data that breaks validation rules and compromises downstream processes.)– Noise in DataThese issues inflate compute costs, introduce bias, and produce unstable predictions, resulting in wasted hours debugging pipelines, increased operational risks, and eroded trust in your AI outputs. Ref.: Data Poisoning: A Silent but Deadly Threat to AI and ML Systems | by Anya Kondamani | nFactor Technologies | Medium As you can see in this example – Despite successfully fetching data, it was unable to render it, due to hitting maximum request limit, as bad data was involved, thus AI Agents face issues if directly brute forced with Raw/Unclean data. But this isn’t just a data science problem. It’s a data engineering problem. And the solution lies in structured, governed data management—beginning with a robust medallion architecture. Ref.: Data Intelligence End-to-End with Azure Databricks and Microsoft Fabric | Microsoft Community Hub Raw → Silver → Gold: The Databricks Way Databricks ML Agents thrive when your data is managed through the medallion architecture, transforming raw chaos into clean, trustworthy features. For those new to Medallion architecture in Databricks, it is a structured data processing framework that progressively improves data quality through bronze (raw), silver (validated), and gold (enriched) layers, enabling scalable and reliable analytics. Raw Layer: Ingest Everything The raw layer is where we land all data, regardless of quality. It’s your unfiltered feed—logs, events, customer input, third-party APIs, CSV dumps, IoT signals, etc. e.g. CloudFronts Case studies as directly fetched from WordPress API This script automates the extraction, transformation, and optional upload of WordPress-based case study content into Azure Blob Storage as well as locally to dbfs of Databricks, making it ready for further processing (e.g., vector databases, AI ingestion, or analytics). What We’ve Done On running the above Raw to Silver Cleaning code in Databricks notebook we get a much formatted, relevant and cleaned fields specific to our requirement In Databricks Unity Catalog → Schema → Tables, it looks something like this: – Silver Layer: Structure and Clean Once ingested, we promote data to the silver layer, where transformation begins. Think of this layer as the data refinery. What happens here: Now we start to analyze trends, infer schemas, and prepare for more active feature generation. Once I have removed noise; I begin to see patterns. This script is part of a data pipeline that transforms unstructured or semi-clean data from a Raw Delta Table (knowledge_base.raw.case_studies) (Actually Silver, as we have already done much of the cleaning part in previous step) into a Gold Delta Table (knowledge_base.gold.case_studies) using Apache Spark. The transformation focuses on HTML cleaning, type parsing, and schema standardization, enabling downstream ML or BI workflow What have we done here ? Start Spark Session Initializes a Spark job using SparkSession, enabling distributed data processing within the Databricks environment. Define UDFs (User-Defined Functions) Read Data from Silver Table Loads structured data from the Delta Table: knowledge_base.raw.case_studies, which acts as the Silver Layer in the medallion architecture. Clean & Normalize Key Columns Write to Gold Table Saves the final transformed DataFrame to the Gold Layer as a Delta Table: knowledge_base.gold.case_studies, using overwrite mode and allowing schema updates. As you can see we have maintained separate schemas for Raw, Silver Gold etc; and at gold we finally managed to get a fully cleaned noiseless data suitable for our requirement. Gold Layer: Ready for ML & BI At the gold layer, data becomes a polished product, curated for specific use cases—ML models, dashboards, reports, or APIs. This is the layer where scale and accuracy become feasible, because it finally has clean, enriched, semantically meaningful data to learn from. Ref.: What is a Medallion Architecture? We can further make it more efficient to fetch with vectorization  Once we reach stage and test again in the Agent Playground, we no longer face the error we had seen previously as now it is easier for the agent to retrieve gold standard vectorized data. Why This Matters for AI at Scale The difference between “good enough” and “state of the art” often hinges on data readiness. Here’s how strong data management impacts real-world outcomes: Without Medallion Architecture With Raw → Silver → Gold Data drift goes undetected Proactive schema monitoring Models degrade silently Continuous feature freshness Expensive debugging cycles Clean lineage via Delta Lake Inconsistent outputs Predictable, testable results (Delta Lake is an open-source storage layer that brings ACID transactions, scalable metadata handling, and unified batch processing to data lakes, enabling reliable analytics on massive datasets.) … Continue reading The Hidden Cost of Bad Data:How Strong Data Management Unlocks Scalable, Accurate AI →