HubSite 365 - The network about Microsoft 365 and Azure

Keynote: Tame Your Rogue AI Agent

HubSite - Editorial — Sun, 12 Jul 2026 00:19:44 +0000

Taming brilliant but unhinged AI agents with GitHub Copilot and Azure AI governance, skills and human judgment

Your Agent is Brilliant, and Completely Unhinged! — In the ColorCloud 2026 keynote, Scott Durow warned that modern AI agents can reason powerfully but also act unpredictably, so teams must design them to be reliable and safe.
Azure AI Foundry introduces core tools to tame agents, including rubric evaluators, the Agent Optimizer, Agent 365 for control, Foundry IQ for knowledge grounding, and a Canvas for custom UIs.
Auto-generated rubrics and self-optimizing agents let teams evaluate behavior automatically, then tune model choice, instructions, tool descriptions, and scaling to improve results over time.
Governance and security rely on Agent 365, which manages identities, access controls, and observability so organizations can audit, monitor, and safely deploy agents in production.
Grounding and interaction use Foundry IQ to anchor answers to trusted documents and data, while the Canvas lets agents build dynamic user interfaces for clearer, context-aware conversations.
Human oversight and production readiness remain essential: keep people in the loop, run continuous evaluations, and treat agents as improving teammates that must be governed and audited before broad deployment.

ChatGPT vs 5 AI Models: Cost Shock

HubSite - Editorial — Sat, 11 Jul 2026 18:37:38 +0000

Copilot Cowork model test reveals credit savings expert tip route to Terra for better Microsoft Copilot results

Copilot Cowork test compared the same prompt across six models (dashboard, PowerPoint, Word, email, ticketing, sentiment). The trial showed wide differences in quality and cost when the teams ran identical tasks.
11.7x cost gap was observed between the most expensive and cheapest model for the same outputs; Terra delivered better cost and Word-doc quality than GPT-5.5, while Opus 4.8 produced the strongest dashboard for the fewest credits.
The practical recommendation: stop routing production Cowork traffic through GPT-5.5 and switch to Terra for similar or better output at lower credit cost.
Microsoft unveiled the MAI family, led by MAI-Thinking-1 (a 35B-active-parameter reasoning model with a 256K context window) and I--1Flash (text-to-code specialist). The company claims roughly 10x lower token costs than comparable third-party models while keeping competitive performance.
The new models use a Mixture of Experts (MoE) approach to reduce inference footprint; Microsoft deploys them on Azure and exposes standard APIs for easy integration into apps and workflows.
Immediate impact: Microsoft already embeds these models in Excel and Outlook to cut third-party fees, and small Aion models can run locally on Windows PCs to enable offline or low-cost use. This shift boosts Microsoft’s strategic independence and lowers enterprise AI costs.

MCP: APIs for LLMs at Scale

HubSite - Editorial — Sat, 11 Jul 2026 01:15:55 +0000

MCP: Reimagining APIs for LLMs with Azure AI and Azure API Management, Copilot demo in a YouTube short

This YouTube Short explains how MCP (Model Context Protocol) acts as an open protocol that unifies how LLMs connect to external tools and data, replacing many custom, model-specific APIs.
The system uses a client-server model where MCP Hosts, MCP Clients, and MCP Servers talk over a standardized JSON-RPC 2.0 schema so agents can discover and use services reliably.
Core capabilities let LLMs call tools, access structured resources (like databases and files), load reusable prompts, and receive real-time notifications, all through one interface.
MCP runs over Streamable HTTP for remote servers or stdio for local servers, and Microsoft provides an MCP C# SDK to simplify integration and model switching for .NET developers.
The protocol boosts cross-vendor interoperability and supports model-agnostic workflows, helps agents adapt to API changes without code rewrites, and can be governed with enterprise security controls on Azure.
Microsoft additions include Azure AI Foundry with native MCP support, an Azure MCP Server preview for natural-language access to services, and Copilot Studio integration to sync tools, prompts, and resources in real time.

Agent Kit: Build Azure Cosmos DB Experts

HubSite - Editorial — Sat, 11 Jul 2026 00:15:55 +0000

Azure Cosmos DB Agent Kit enforces best practices, prevents partition key leaks, optimizes queries and SDK usage

Azure Cosmos DB Agent Kit turns AI coding assistants into Cosmos DB experts with a one-command install.
It activates automatically when you work on Cosmos DB code so developers get guidance without leaving their editor.
The kit enforces a large, curated rule set: 120+ rules organized into 12 categories.
Rules cover data modeling, partitioning, indexing, query patterns, vector and full-text search, and developer tooling.
It prevents costly production mistakes by flagging issues like a missing partition key filter or unsafe queries.
Examples in the demo include fixing a fan-out query and stopping tenant data leakage in a multi-tenant app.
SDK best practices and query recommendations help reduce request unit (RU) costs and improve performance.
Advice includes correct client initialization, retry logic, indexing choices, and practical query optimization tips.
The kit works with common AI pair-programmers and requires Node.js with npm/npx to install (it runs via the skills format).
Installation is a single command and the skill integrates into editors and CLIs that support Agent Skills.
Microsoft validated the release before GA with extensive testing: General Availability (GA) follows > 200+ automated tests.
It also guides infrastructure as code and deployment patterns for Terraform, Bicep, and container setups.

Excel Custom Skills: How They Work

HubSite - Editorial — Fri, 10 Jul 2026 19:15:42 +0000

Microsoft guide to Excel Custom Skills and Copilot in Excel with Power BI training for VLOOKUP to XLOOKUP and formatting

Excel Custom Skills are user-created instruction sets that teach Copilot how to perform repeatable tasks in Excel.
They live in your OneDrive as a folder with a main file named SKILL.md, so Copilot can find and run them automatically.
Each skill file uses a simple structure: a Frontmatter (YAML) section for metadata and a Markdown body with step‑by‑step instructions.
This lets you define when the skill should trigger and exactly how the task should be completed.
Custom Skills boost productivity by providing Automation and Consistency across repeated workflows like invoice reviews, report formats, or data conversions.
They also give Personalization so the AI follows your team’s rules and tone.
To create a skill: set up the /Documents/Cowork/skills/ folder in OneDrive, add a subfolder named with lowercase hyphens, then add a SKILL.md file and any companion files.
Keep the skill name in the file’s frontmatter identical to the folder name.
Use skills by asking Copilot or running the skill directly; you can Run, Undo, Refresh, and Activate them as needed.
Copilot recognizes skills when the workbook context and description match the skill’s metadata.
Limits and troubleshooting: Copilot Cowork supports up to 50 skills per account, each up to 1 MB, with up to 20 companion files.
Common fixes include correcting the frontmatter, matching folder/name conventions, and reloading skills if Copilot does not detect changes; sample skills (like a calendar demo) help test behavior.

Power Platform: Automate Control Charts

HubSite - Editorial — Fri, 10 Jul 2026 18:15:42 +0000

Microsoft expert: Automate manufacturing control charts with Power Platform Dataverse Copilot Power BI and Power Automate

Power Platform automates shop-floor quality by generating and updating control charts in real time.
It captures data, analyzes trends, and shows results on dashboards so teams see issues fast.
Core components: Power Automate moves data, Power BI visualizes charts, Power Apps lets operators log and act, and Dataverse stores trusted quality records.
These parts work together to remove manual steps and centralize data.
SPC workflow: collect process data, compute control limits (mean, UCL, LCL), plot points, detect out-of-control signals (e.g., Western Electric/Nelson Rules), and trigger alerts and corrective actions automatically.
Teams can re-baseline processes after confirmed changes.
Key benefits: delivers real-time visibility, reduces manual errors and data entry, enables automated statistical analysis, and speeds up response to defects.
This supports compliance and helps reduce downtime.
Integration & data flow: the solution pulls data from PLCs, IoT sensors, MES/ERP systems, or manual inputs into Dataverse.
That creates a closed-loop quality system where alerts drive root-cause work and corrective actions.
Practical outcomes: operators use apps on the floor, managers see embedded charts in dashboards, and teams run targeted investigations when outliers appear.
The approach scales into existing reporting and improves accuracy and production performance.

Azure Update 10th July 2026

HubSite - Editorial — Fri, 10 Jul 2026 17:15:42 +0000

Azure update: OpenShift region, WAF, Blob SFTP Entra ID, Event Hubs security, Chaos Studio, OpenAI GPT, AVD

Azure Foundry now hosts large models like Claude Sonnet 5 and supports open-weight models such as Kimi K2.7 to run inside Microsoft’s governance boundary.
These integrations keep data and model execution on Azure and improve coding and agent capabilities for developers.
Client-side CRC64-NVME adds byte-level integrity checks for Azure Blob Storage to ensure data written is exactly what was sent.
Use this to detect corruption early and to strengthen end-to-end data reliability during transfers.
Azure Storage Mover gained support for migrating from Google Cloud Storage using S3-compatible endpoints and private network options.
This simplifies moving large buckets to Azure Blob Storage without exposing data to the public internet.
Entra ID SFTP replaces legacy SFTP access for Blob Storage, centralizing authentication and improving auditability.
At the same time, the Quantum Safe Program advances TLS 1.3 and modern cryptography to future-proof data protection.
Event Hubs Dedicated Confidential Computing and Network Security Perimeter support protect high-throughput streaming with hardware-backed isolation and tighter network controls.
These features suit regulated or high-security real-time ingestion scenarios.
Developer and security controls include WAF rule exceptions (Preview), a GA PostgreSQL PowerShell module, and new Chaos Studio workspaces and CLI tooling.
Together they improve operational flexibility, testing, and automation for cloud apps and infrastructure.

AI That Transforms Your Business

HubSite - Editorial — Fri, 10 Jul 2026 16:15:42 +0000

How SharePoint.md powers Copilot and always-load Skills for governance, branded dashboards in Microsoft three sixty five

SharePoint.md: A central file that holds your business terms, brand rules, library maps, naming conventions, and always-load instructions so Copilot reads the same context every session.
Copilot Skills: Small skill files that run alongside SharePoint.md and can be auto-loaded at session start to provide consistent capabilities and actions for every user.
Governance as a file: Governance moves from training people to maintaining a readable file that enforces rules and limits Copilot will never break when generating outputs.
Always-load rules: Rules that trigger Skills and context automatically at sign-in, ensuring consistent brand voice, compliance, and naming across the site.
Copilot updates itself: Demonstrates that Copilot can modify or add Skill files (for example, “remember from next quarter”), letting the system adapt without manual reconfiguration.
Business impact: Standardized context and agent-driven automation speed decisions, cut task time, improve accuracy, and deliver measurable productivity and cost benefits.

Microsoft AI: Copilot/Cowork/Scout/MCP

HubSite - Editorial — Fri, 10 Jul 2026 16:15:42 +0000

Microsoft decodes Copilot Cowork Scout MCP Agent Builder and Copilot Studio to clarify AI roles security and usage

Ladder Framework decodes Microsoft’s confusing AI names into one clear model for decision-making.
It maps six similar-sounding products so you know which tool fits each real-world task.

Microsoft Scout is an always-on Autopilot agent that runs on Windows 11 and can access local files, run code, and execute shell commands under admin guardrails.
It works proactively (not just by prompt) and supports human review for high-risk actions.

Copilot Cowork is a cloud-only agent for multi-step workflows across Microsoft 365 apps like Teams, Outlook, and SharePoint.
It handles scheduled or orchestrated tasks but cannot modify local files or run local commands.

Copilot remains the interactive assistant for immediate needs—drafting emails, summarizing docs, and searching across work content.
Use it when you need quick, on-demand help inside apps.

Copilot Studio (low-code) and Agent Builder (no-code) let teams create custom agents and workflows without starting from scratch.
Studio targets developers with more control; Agent Builder targets non-developers for fast agent creation.

MCP (Model Context Protocol) connects Copilot and agents to external data sources (APIs, databases, file systems) so agents work with live data.
Agents use human-in-the-loop controls and GitHub Copilot Business or Enterprise credits for licensing, which helps manage cost and security.

Microsoft Fabric: AI Reinvents Power BI

HubSite - Editorial — Fri, 10 Jul 2026 13:13:40 +0000

Microsoft expert explores Power BIs future in Fabric, covering Copilot, Semantic Views, Fabric web app and prompt-first UI

Power BI future overview: Mohammad Ali explains where Power BI is heading and how it fits inside Microsoft Fabric.
He highlights recent shifts toward AI, web-first workflows, and tighter integration across analytics tools.
Semantic Views: a shared metadata layer that makes models consistent and reusable across teams.
Semantic Views simplify report building, speed up analysis, and improve governance by centralizing business logic.
Fabric Web App and Direct Lake: the browser-first experience brings more work into the web with faster live queries from OneLake.
Direct Lake reads Delta tables directly so reports feel as fast as imported data without duplicating it.
Copilot in Power BI: conversational AI that generates visuals, DAX, and narrative summaries from natural-language prompts.
Copilot-for-visualization builds charts from descriptions, while reusable AI skills support repeatable transformations and features inside reports.
Power BI Desktop updates and prompt-first ideas: Desktop gains tighter live editing and sync with the service, plus better history and collaboration.
The team is exploring prompt-first interfaces, but will keep power-user capabilities while adding natural-language workflows.
Version history and collaboration: OneDrive-based versioning, Git integration, and Task Flows improve teamwork and deployment.
Automatic mobile layouts and service updates make reports easier to share, govern, and maintain across organizations.

Power BI Card Visual Leaves Preview

HubSite - Editorial — Fri, 10 Jul 2026 12:13:40 +0000

Microsoft Power BI Card visual out of preview, modern multi-measure cards boost KPI design and report performance

The New Card visual is now generally available and replaces the legacy Card and Multi-row Card as the default Card in Power BI.
Use it to display KPIs and summary values with a modern, unified tool.
The visual supports multiple measures in a single container, plus category grouping and small multiples for side-by-side comparisons.
This lets you show several metrics together without stacking many separate visuals.
Use advanced formatting to polish callouts: per-card styles, images or icons, shapes, shadows, accent bars, dividers, and customizable padding.
These options give fine control over look and emphasis for each metric.
Consolidating metrics into one visual improves performance and report design.
Fewer visuals mean faster rendering, simpler maintenance, and cleaner layouts.
You can restore the legacy card for comparison, but migrate reports to the new visual to simplify upkeep and access modern features.
Update legacy card sections to reduce clutter and improve consistency.
Quick start: add the card from the Visualizations pane, drop in measures and category fields, and choose Card or Table style.
Note the new default padding (12px) and responsive rendering so content scales nicely when resized.

Power Query: Advanced Options Debunked

HubSite - Editorial — Fri, 10 Jul 2026 06:31:37 +0000

Microsoft Excel Power Query expert tips to preserve manual edits on refresh, streamline data transforms and automation

Power Query: A data-preparation tool inside Excel and Power BI that cleans, reshapes, and merges data using a visual interface and the underlying M code language.
It helps non‑developers automate repeatable data tasks and connect to many sources.
Advanced Options Limitations: Many “Advanced” settings are partially implemented or conditional, not fully flexible.
Examples include Pivot Column failing with “Don’t aggregate,” Replace Values showing extra options only after converting to text, and some connector features available on desktop but not online.
Advanced Editor: The Advanced Editor exposes full M code so you can edit steps directly, create functions, and work around GUI limits.
Use it when the interface hides options or when you need precise, repeatable transformations.
Refresh and Manual Edits: A query refresh replaces workbook data and can overwrite manual changes.
To preserve manual edits, keep user changes separate from query outputs or build queries that merge new data with manually maintained tables before loading.
Performance Best Practices: Improve speed by filtering early, disabling unnecessary type detection, and minimizing transformation steps.
Push heavy work back to the data source when possible to reduce local processing time.
When to Use GUI vs Code: Use the GUI for common tasks like grouping, pivots, and simple replace operations for faster development.
Switch to the Advanced Editor when you need unsupported options, conditional logic, or more efficient, repeatable solutions.

Copilot: Automate Tasks, Not Teams

HubSite - Editorial — Fri, 10 Jul 2026 04:12:43 +0000

Microsoft expert: Use Copilot, Power Platform and Copilot Studio to automate low risk accounts payable with Purview

Episode overview: This YouTube episode of The AI Frontier Playbook features Eric Murray, Director of Innovation at Kezber, explaining how to automate sales and back-office work using Copilot, Power Platform, and Copilot Studio.
He focuses on practical, low-risk steps teams can deploy without overwhelming staff.
Start small and practical: Avoid automating an entire department at once; pick repetitive, low-risk processes owned by a reliable champion who knows every exception.
This approach delivers value fast and lowers the chance of project failure.
Design: split deterministic work from AI reasoning — use plain automation for rule-based steps and AI only where judgment or unstructured data matters.
Eric walks through an accounts payable example, slicing invoice flows into small chunks that move data from mailbox to ERP reliably.
Human oversight and trust: Build a Human-in-the-loop model so AI drafts or suggests and people approve exceptions, which increases user trust and reduces workarounds.
Design clear escalation paths and transparent feedback loops so humans stay in control.
Security and compliance: Treat security and data residency as core work — include access controls, governance, and regional privacy rules (e.g., Quebec Law 25) before scaling.
Policy, auditing, and early governance choices prevent costly rewrites later.
Measure and sustain value: Track ROI, user adoption, and model costs after launch; monitor models to avoid “solution rot” and optimize spending (Eric cites dramatic savings from a single model swap).
Diagnose issues as tool gaps or adoption problems and iterate in weeks, not months.

Power BI: Expert Gone — Your Next Steps

HubSite - Editorial — Fri, 10 Jul 2026 03:12:43 +0000

Microsoft expert reveals how Copilot for Power BI and Fabric accelerate semantic model understanding, DAX, and governance

Power BI semantic model — The video shows that most business logic, calculations, and assumptions live inside the semantic model and often remain only in the original developer’s head.
Copilot helps teams explore that hidden knowledge when the original author is unavailable.
Copilot for Power BI — Use Copilot to explain complex measures, show how revenue and KPIs are calculated, and point out which parts of the model matter for a given business question.
It speeds model understanding without requiring the original author.
Measure documentation — Copilot can find undocumented measures, generate clear descriptions, and recommend formatting and naming conventions.
That makes measures easier to maintain and share with others.
Data governance — The walkthrough highlights how Copilot improves model maintainability and governance by capturing business logic and context directly in the model.
This reduces single-person dependencies and supports safer changes in production.
Model validation — Before writing new DAX, Copilot helps validate whether the existing model can answer new questions, avoiding unnecessary code and duplicate measures.
Use it to check feasibility and identify gaps first.
Production-ready measures — The video demonstrates creating, reviewing, and validating DAX with Copilot, then adding descriptions and formatting so measures are ready for production.
Follow this practice to keep dashboards reliable and easier to hand over.

Excel AI: Surprising Test Results

HubSite - Editorial — Fri, 10 Jul 2026 02:12:43 +0000

Microsoft Excel AI meets Tracelight to turn PDF ledgers into automated financial models and live forecasts with Power BI

Tracelight: The hands-on video shows Tracelight importing a messy 12‑month PDF ledger into Excel and building a linked P&L and Balance Sheet in minutes.
It traces every number back to underlying transactions so analysts can verify sources quickly.
Data extraction & reconciliation: Tracelight maps accounts, classifies transactions, and reconciles the Balance Sheet automatically, cutting hours of manual work.
The tool also lets you save and reuse prompts to speed repeated workflows.
Interactive forecasting: The demo converts finished models into a live forecast app where anyone can change assumptions and see instant results.
It highlights version differences and spots hidden errors before they affect decisions.
Agent Mode: Microsoft Excel’s Agent Mode acts like an autonomous coworker that plans and runs multi‑step tasks such as cleaning data and building dashboards in real time.
These agents reduce repetitive work and scale complex workflows across sheets.
=COPILOT(): The new in‑cell assistant lets you ask plain‑English questions and returns charts, summaries, or cell transformations without writing complex formulas.
This helps non‑technical users extract insights and speeds routine analysis.
AI formula completion: Excel now predicts and completes formulas to cut syntax errors and speed model building, while Copilot tools assist with data cleaning and visual storytelling.
These features require a Microsoft 365 Copilot subscription and can use different AI models for specific logic or performance needs.

Power BI: From Reports to Fabric Apps

HubSite - Editorial — Fri, 10 Jul 2026 01:12:43 +0000

Microsoft expert guide to moving Power BI reports to Microsoft Fabric apps with Copilot and design best practices

Fabric Apps let teams build full web applications inside Microsoft Fabric instead of just static dashboards.
They shift from drag-and-drop reports to code-first, browser-hosted apps that run like web pages.
Visualization as code uses TypeScript and libraries like Vega or D3 to create custom visuals and interactions.
Developers use the Rayfin SDK to define models, UI, and behavior in code for precise control.
Built-in backend features include authentication, session handling, hosting, and automatically generated GraphQL APIs.
These services reduce boilerplate so teams can focus on app logic and user experience.
OneLake integration gives apps direct access to Fabric data and managed storage.
Apps can read and write to Fabric databases without custom infrastructure setup.
Write-back enables users to update data from the app interface with version tracking and controls.
This makes Fabric Apps useful for planning, workflows, and interactive tools that need real-time edits.
Choose Fabric Apps for rapid prototyping, deep customization, and internal tools that need action triggers and writable data.
Keep using Power BI reports for fast, no-code dashboards; use Fabric Apps when you need full app behavior and custom visuals.

Azure Health Model Explained

HubSite - Editorial — Fri, 10 Jul 2026 00:12:43 +0000

Microsoft Azure Health Models and Copilot Cowork billing with Azure Monitor and Cost Management for planning and control

Azure Monitor Health Models turn raw telemetry (metrics, logs, traces, Prometheus) into a single, business-focused health view that answers “Is my app healthy?”.
Key building blocks are Service Groups (group related resources), Entities (components like VMs or APIs) and defined Health States that roll up component status into an overall application rating.
Use the Graph view for real-time, color-coded dependency maps and the Timeline view to review historical trends and spot recurring issues for faster root-cause analysis.
Primary benefits: cut alert fatigue by surfacing only business-impacting incidents, speed troubleshooting with visual dependency mapping, and enable automation driven by health signals (auto-scale, failover, runbooks).
Operational guidance: map critical user journeys, set baselines and health thresholds, create health-driven alerts, and integrate models with incident playbooks to plan, control, and automate responses.
For usage-based billing like Copilot Cowork, estimate costs by mapping usage to workloads, set quotas and budgets, monitor cost alongside health, and use cost observability to prioritize actions that reduce spend while protecting customer experience.

PowerPoint: Reveal Timeline with Morph

HubSite - Editorial — Thu, 09 Jul 2026 22:12:43 +0000

Microsoft PowerPoint expert reveals morph timeline for medical presentations to reveal stages and sharpen audience focus

Morph transition and timeline reveal: Use PowerPoint’s Morph to animate movement, size, color, and cropping so each timeline stage appears focused while others fade into the background.
This creates a smooth, cinematic flow that guides the audience step-by-step.
Prepare base image: Start with a high-resolution timeline image, apply grayscale and blur to non-focused areas, then crop to show only the first stage on slide one.
Keep the image aligned and centered so Morph can match objects across slides.
Duplicate slides and sequence the reveal: Duplicate the slide, adjust the crop to show the next stage, and let Morph animate the change between each slide.
Repeat for each stage to build a smooth progressive narrative.
Timing and control: Set transition duration (commonly ~1 second), then choose manual clicks or automatic timing depending on live or self-running presentations.
Use consistent timing to maintain rhythm and avoid distracting jumps.
Benefits for medical talks: This technique improves audience engagement and visual clarity, reducing cognitive overload when explaining processes like disease progression or treatment stages.
It makes dense data easier to follow without extra software.
Compatibility and reuse: Works in PowerPoint 2016+ and Microsoft 365 using Picture Format, Artistic Effects, Crop, and Morph; build a reusable master slide to adapt timelines quickly for future talks.
This keeps your presentations consistent and easy to update.

Power Platform: Choose External Models

HubSite - Editorial — Thu, 09 Jul 2026 00:07:15 +0000

Microsoft expert: Power Platform Shorts on selecting external AI models for Copilot and Azure OpenAI in a YouTube Short

External model selection
Quick summary of the YouTube Short: choose the right third‑party or cloud AI model to power Power Platform experiences. Match model strengths to your scenario (text, summarization, Q&A) and plan integration steps before building.
Connectors & endpoints
Use prebuilt or custom connectors to call model APIs. Configure endpoints, map inputs/outputs, and test responses in a sandbox before production.
Authentication & security
Protect keys and tokens with Azure AD and Azure Key Vault. Limit what data you send to external models and apply Power Platform DLP rules to prevent leaks.
Performance & cost
Compare latency, throughput, and pricing for candidate models. Reduce cost with batching, caching, and fallback logic when models are slow or unavailable.
Quality & prompting
Tune prompts and use templates to improve answers. Monitor for hallucinations, validate outputs, and add post‑processing checks where accuracy matters.
Governance & monitoring
Log usage and errors, set quotas, and audit calls for compliance. Build telemetry and alerts so you can measure impact and roll back or switch models if needed.

Microsoft Copilot Update: Key Changes

HubSite - Editorial — Wed, 08 Jul 2026 18:25:14 +0000

Microsoft Copilot updates streamline Microsoft 365 with Copilot Cowork, Loop, Copilot Chat and agentic mode

Overview: Microsoft updated Copilot’s interface and features to simplify and standardize the experience.
These changes rolled out in June 2026 with wider availability expected by August 1, 2026.
New UI consolidation: A refreshed Waffle icon groups Notebooks and Create Agent together and the main Copilot icon moved inside Microsoft 365 apps.
The chat entry point now sits at the top-left to make access more consistent across apps.
Agentic mode & task visibility: Copilot’s new Agentic mode runs longer, multi-step tasks autonomously inside 365 apps.
Long-running jobs now show taskbar progress and status in Windows so users monitor workflows without opening each app.
Governance and content integrity: AI-generated audio and video can carry organizational Watermarks to flag altered media.
Copilot also applies the highest Sensitivity labeling from source content and warns users before sharing if labeling fails.
Prompt and creative tools: Copilot adds a new Prompt Gallery and Prompt Lab for reusable prompts and experiments.
PowerPoint image generation is faster with the MAI-Image-2-Efficient model for quicker visual iterations.
Search, answers and skills: Copilot Search now finds people by department and Copilot Chat returns rich answer cards (weather, stocks, structured data) instead of plain text.
AI skill inferencing expanded to Microsoft 365 E3 and E5 plans for broader talent and workflow insights.

SharePoint QMS: Fixes for ISO 9001 Fails

HubSite - Editorial — Wed, 08 Jul 2026 13:05:23 +0000

Microsoft expert on SharePoint QMS remedies document control, versioning and training to stay audit ready and compliant

ISO 9001 audits usually fail because organizations have execution gaps, not just missing papers.
The video shows that poor tracking and weak processes cause most findings, so fix the system, not only the files.
Calibration records often lack dates, traces to standards, or scheduled renewals.
Auditors expect clear traceability, so track equipment, due dates, and evidence consistently.
CAPA processes fail when teams close actions without real fixes.
Use strong root cause analysis to stop repeats and show durable corrective or preventive steps.
Document control breaks down when users can’t prove the current file or approver.
Enforce a clear version history, approval steps, and document lifecycles so auditors see the single accepted source.
Training records are a common gap when staff competence is not linked to roles or processes.
Keep a visible competence tracker that ties courses and evidence directly to job roles and procedures.
SharePoint QMS fixes many problems by centralizing documents, permissions, and history.
Implement automated workflows, metadata, and audit trails to enforce reviews, approvals, and instant evidence during audits.

Power BI: Build Custom DAX Functions

HubSite - Editorial — Wed, 08 Jul 2026 12:05:23 +0000

Power BI DAX UDFs unlock reusable DAX functions to streamline models and reuse logic across measures and reports

DAX UDFs let you package reusable, parameterized DAX logic so you can define a calculation once and use it across measures, columns, tables, and reports.
They reduce copy-paste, keep results consistent, and make models easier to maintain.
Turn on the preview preview feature in Power BI or Fabric to access UDFs, then write functions inside DAX Query View where the new function editor appears.
This view helps you author, test, and manage functions in one place.
Declare functions with the new FUNCTION / DEFINE FUNCTION syntax and supply parameters for scalar, table, or reference inputs.
Parameters make the logic flexible and let you reuse the same function for different scenarios.
Test functions using EVALUATE to inspect results before updating the model, then call the custom function from a normal measure when ready.
This workflow lets you validate logic independently and avoid breaking reports.
Add clear descriptions and metadata so functions appear with helpful details in the DAX experience, and store them in the model object file functions.tmdl for team sharing.
Use type hints and modular design as best practices, and keep function code under version control to track changes and promote reuse across projects.
Well-documented, parameterized UDFs make your analytics more reliable and easier to scale.

Dynamics 365 Data Exploration Agent Demo

HubSite - Editorial — Wed, 08 Jul 2026 06:23:12 +0000

Microsoft expert demo of Dynamics Three Sixty Five Data Exploration Agent in a YouTube Short with Copilot AI insights

Data Exploration Agent is an AI-powered search tool for model-driven apps like Dynamics 365 and Power Apps that lets users find and filter records using plain language.
It removes the need to build complex advanced filters manually.
The agent uses natural language processing (NLP) and Copilot AI to interpret queries such as “show accounts updated in the last month” and return the matching records instantly.
Users type intent and the system translates it into precise filters.
Key features include a preview “Smart Grid” that accepts natural-language queries and a new “Visualize with Copilot” option that creates charts from filtered data on demand.
Visual charts update interactively so users can refine results inside the chart view.
Benefits include faster data access, higher productivity, and broader accessibility for non-technical staff who don’t know advanced filter syntax.
Teams spend less time building views and more time analyzing insights.
Microsoft announced the public preview on February 18, 2025, and it is part of the 2024 Release Wave 2, rolling out via the Monthly Channel.
The feature is in preview and requires admin opt-in.
To enable it, an admin opens the Power Apps environment, finds the Natural Language Grid and View Search setting, turns on the manual opt-in, and then tests the agent in a model-driven app.
This setup step lets teams try the tool before broad deployment.

SharePoint Version History: Cut Storage

HubSite - Editorial — Wed, 08 Jul 2026 02:20:21 +0000

Microsoft expert tips to manage SharePoint version history, reduce storage waste and control versions in libraries

SharePoint Version History: A built-in feature that saves every saved state of a file so teams can recover edits, track changes, and support co-authoring.
It keeps a timeline of a document’s lifecycle and helps with audits and accidental restores.
Default 500 versions: By default SharePoint can retain up to 500 versions per file with no expiry, which can dramatically increase storage use.
Each version is a full copy, so a single large file can multiply into a huge storage footprint.
Major vs minor versions: Libraries can store full major versions and optional minor (draft) versions; both count toward storage.
Autosave and frequent co-authoring often create many versions quickly, especially for large files like videos or drawings.
Version settings: Admins can manage versions using three main options—an automated cleanup mode, a manual fixed-number limit, or a manual limit with a time-based expiry.
Choosing the right mode lets you balance recovery needs against storage costs.
Storage fixes: Reduce storage by setting sensible version limits, enabling major-only history where appropriate, deleting old versions, and applying retention policies.
Monitor libraries, restrict versioning for large file types, and use admin tools or PowerShell to audit and enforce settings.
Best practices: Define versioning policies by file type and team needs, train users on saving behavior, and run regular storage audits.
These steps keep document libraries efficient while preserving the ability to recover important work.

Rayfin Data App: Power BI + TypeScript

HubSite - Editorial — Wed, 08 Jul 2026 01:20:21 +0000

Rayfin Data App redefines Power BI on Fabric with TypeScript/React Semantic Model and GitHub Copilot for cost savings

This video demonstrates a new approach to Power BI delivery using the Rayfin Data App template that combines a TypeScript/React frontend with a Power BI Semantic Model backend.
The presenter shows how this shifts reporting from static PBIX files to full web applications.
The app architecture uses Microsoft Fabric Apps features: managed SQL, GraphQL/REST endpoints, and static hosting are provisioned for you.
Rayfin works with Fabric to remove manual infrastructure setup and speed development.
The video explains cost gains: host the app on a low-cost F2 Fabric capacity and share with free Fabric users instead of buying a Power BI Pro license per viewer.
For large audiences this reduces licensing costs and improves user reach.
Development becomes code-first: the Rayfin CLI and TypeScript model classes auto-generate APIs and database schemas.
The demo also shows using GitHub Copilot prompts to speed enhancements without manual coding.
Security and governance stay centralized: the Semantic Model remains the trusted data layer and the apps use Microsoft Entra ID for authentication.
This avoids data duplication and removes the need for embed tokens.
Practical impact: teams can build custom, type-safe apps with richer UX while keeping a single source of truth.
The approach scales distribution, preserves governance, and leverages existing Power BI investments.

M365 Message Center: Top Updates Ep 432

HubSite - Editorial — Wed, 08 Jul 2026 00:20:21 +0000

Microsoft Message Center highlights Power Automate flow recovery, OneDrive library browsing, Teams Facilitator, Copilot

Message Center and AI-driven message tracking: This episode shows a chat-based Message Center that lets admins ask for and retrieve posts relevant to their tenant.
It speeds up change triage and reduces manual searching.
Copilot and service plan behavior: Copilot now helps create SharePoint pages, generate charts, and suggest Excel formulas, while Microsoft updated Copilot service plan behavior for deployments.
These changes speed content creation and clarify licensing expectations.
Microsoft Teams Recap and Facilitator: The Recap app centralizes meeting summaries in one place, and Facilitator detects and helps resolve knowledge gaps during meetings.
Both features improve follow-up and team alignment.
Power Automate and restore deleted flows: You can now recover accidentally deleted flows to reduce downtime and rework.
Check retention details and restore steps to ensure reliable process recovery.
OneDrive libraries and brand assets in PowerPoint: OneDrive for the web offers a new libraries browsing experience, and PowerPoint can surface brand assets for consistent slide design.
These updates streamline content access and maintain design consistency.
Standardized headers and admin scripts: Message Center posts now use headings like "What and Why" and "Rollout Schedule"; update admin scripts to parse the new format.
Adopt agentic workflows to automate reactions to Message Center updates.

Excel AI Skills: The Feature You Need

HubSite - Editorial — Tue, 07 Jul 2026 22:20:21 +0000

Microsoft expert on Excel AI skills with Copilot in Microsoft three sixty five automates tasks and boosts productivity

Copilot Skills: a library of pre-built or custom workflows inside Excel that automate multi-step tasks so the app performs actions for you instead of only giving suggestions.
Plan mode: Copilot shows a step-by-step plan of changes (for example, "change formula in cell B5") and asks you to approve or edit before it makes any edits, so you keep control and can review results easily.
Creating and sharing skills: build custom skills, save them as a SKILL.md file in OneDrive, and share them across teams; custom skills are in early preview via the Insider channel with broader rollout expected soon.
Key benefits: cuts repetitive work, reduces errors, standardizes outputs for easier review, and helps people with less Excel experience run complex analyses reliably.
Data connections and automation: Copilot Skills can pull live external data from finance platforms and run workflows automatically, enabling richer reports and reducing manual data entry; this approach may replace many traditional macros over time.
Common use cases: building discounted cash flow (DCF) models, variance analysis, refreshing monthly dashboards, and creating invoice reports—skills make these repeatable and easy to reuse across projects.

ChatGPT: Dont Upload Excel—Try This

HubSite - Editorial — Tue, 07 Jul 2026 21:20:21 +0000

Dont upload Microsoft Excel to ChatGPT—use Excel Power Pivot, DAX and Copilot guardrails to validate joins and margins

The video urges you to stop uploading files and introduces ChatGPT for Excel powered by GPT-5.5, a native add-in that puts AI directly inside Excel for live analysis.
By working with a live workbook, the AI reads cell references, formulas, and ranges so results update automatically without re-uploading files.
Embedding the tool inside Microsoft improves data security, lowers the risk of leaks, and aligns with enterprise compliance and governance rules.
The add-in enables direct formula integration, letting the AI build, fix, or optimize formulas and preserve your existing model structure.
The presenter recommends a guardrail prompt and shows how to inspect the workbook first for missing values, duplicate rows, and hidden sheets before deeper analysis.
The demo runs practical checks—table joins, profit-margin math, and a hallucination test—and stresses verifying outputs with quick checks or Power Pivot cross-checks.

Power Apps Date Tips: EDate and EOMonth

HubSite - Editorial — Tue, 07 Jul 2026 20:20:20 +0000

Microsoft Power Apps tip: EDate and EOMonth in PowerFX with Sequence AddColumns and DateValue to fix end of month dates

EDate: Adds or subtracts months while keeping the same calendar day when possible.
If the target month has fewer days, EDate returns that month’s last valid day (for example, January 31 → February 28).
EOMonth: Returns the last day of the month after adding or subtracting months.
Use EOMonth(..., 0) to get the current month’s last day without extra logic.
Sequence and AddColumns: Use Sequence to create a list of rows and AddColumns to compute dates for each row.
Combine them to generate a run of monthly dates (for schedules, reports, or renewals).
DateValue vs Date: Prefer DateValue when your input is text because it reliably parses date strings.
Use Date for building dates from numeric parts (year, month, day).
Use cases: Ideal for subscriptions, billing cycles, maturity dates, and month-end reporting.
EDate keeps the same day each month; EOMonth forces month-end, reducing manual errors for different month lengths.
Practical tips: Test edge cases like months with 28, 29, 30, and 31 days.
Store results as date-only values when time is irrelevant, and prefer simple formulas like EDate(date, 1) for next-month recurrence or EOMonth(date, 1) for next month’s end.

Microsoft Education: Summer 2026 Updates

HubSite - Editorial — Tue, 07 Jul 2026 06:18:53 +0000

Microsoft summer AI update for educators and students: Copilot family, Teams Assignments, Learning Zone, Windows eleven

Student-centered AI in Microsoft Education shifts tools from answer machines to interactive learning coaches that respect privacy and teacher guidance.
Schools now focus on responsible, educator-informed AI that works inside existing Microsoft 365 workflows.
Teachers can create standards-aligned Unit Plans fast using Teach in the Microsoft 365 Copilot module; the tool supports many countries and produces overviews, weekly lessons, and essential questions from a few inputs.
New Student AI Guidelines let teachers set allowed AI use per assignment (from no AI to full use of Copilot Chat) and use Learning Groups to adapt expectations for different students.
Learning Zone and the Learn app for Windows 11 give teachers live, educator-paced lessons with near-real-time student activity views and direct integration into assignments.
The Study and Learn Agent (Copilot Chat for ages 13+) acts as an interactive tutor that guides practice and gives feedback, while Copilot Notebooks convert class content into structured, interactive study guides at no extra cost for Microsoft 365 Education.
New Learning Activities (Fill in the Blanks, Matching, Quizzes) will integrate into Teams Assignments and supported LMS platforms via the M365 LTI app, making practice more interactive inside existing workflows.

Copilot Studio: Build ROI-Driven Agents

HubSite - Editorial — Tue, 07 Jul 2026 00:37:14 +0000

Microsoft expert: Architect Copilot Studio agents that drive ROI using Power Platform, Dataverse and Power Automate

Video summary: This presentation shows how to design AI agents in Microsoft Copilot Studio that prove business value.
It focuses on measurable ROI and real-world deployment steps for enterprise teams.
Proven ROI: Choose use cases that clearly save time or money and measure results.
Track savings either per-run (time/cost per task) or per-tool (estimates for each tool an agent uses) to quantify impact.
Computer-using agents: Agents can operate UI elements, automate web and desktop tasks, and combine API and UI actions.
Use these agents to automate multi-step workflows that legacy APIs alone cannot handle.
Structured flows: Design predictable conversations and decision paths with visual orchestration and agent nodes.
Structure reduces errors, makes behavior repeatable, and helps teams validate outcomes before rollout.
Agent-to-agent: Enable agents to communicate and delegate tasks to each other for complex automation.
This improves scalability and allows specialized agents to collaborate on larger business processes.
Production readiness: Test agents, monitor analytics, and iterate based on measured performance.
Start with high-impact, low-risk pilots, focus on measurable outcomes, and refine agents until they are reliable and predictable.

PowerPoint: 3 Steps to Declutter Slides

HubSite - Editorial — Mon, 06 Jul 2026 20:06:03 +0000

Microsoft PowerPoint expert tips to declutter complex slides with a three step framework for clarity and readability

3-step decluttering method
Video explains a simple three-step method to clean up complex PowerPoint slides and shows real before-and-after makeovers to demonstrate each step.
Focus on key points
Keep only essential information on each slide, use short phrases or bullets, and move supporting details into speaker notes to avoid text overload.
Streamline visuals
Use clear charts and images, a consistent color and font scheme, and ample white space to make data easier to read and compare.
Create strong hierarchy
Make headings larger and bolder, align elements consistently, and group related items so the audience can scan slides quickly.
Organize and reduce distractions
Use slide masters and sections, limit animations and template clutter, and add light borders or spacing to separate content blocks for better readability.
Clear benefits
Decluttering improves audience engagement, speeds understanding of complex data, and gives your presentation a more professional, confident look.

Claude Skill Powers SharePoint App

HubSite - Editorial — Mon, 06 Jul 2026 19:06:03 +0000

Ported Claude Skill into SharePoint with Microsoft Lists and Copilot proving Skills portability and creative prompts

Hands-on test: I downloaded Anthropic’s Claude Skill, dropped it into a SharePoint Agent Assets library, and ran it against a Microsoft List holding a conference itinerary.
It produced a live, interactive event page (confetti, ticker, filters, tabs, modals) and a second run produced a different creative interpretation from the same input.
Shared architecture: Claude Skills and SharePoint Skills use the same SKILL.md architecture with identical YAML frontmatter, progressive disclosure, and reference patterns.
One skill file can work on both platforms without reauthoring.
Integration layer: The connection relies on the Model Context Protocol (MCP) to let Claude read and act on content across Word, Outlook, Teams, SharePoint, and OneDrive.
That protocol enables a multi-model, agentic workflow rather than a simple chat overlay.
Security model: Claude respects normal Microsoft permissions via Microsoft Entra, using a read-only, delegated permission model so the AI only accesses data the user is allowed to see.
This preserves tenant security while enabling direct content queries.
Operational benefits: Agent Skills let teams store SOPs as a SOP as Markdown file and have the AI follow them consistently, removing repetitive setup and manual file uploads.
Progressive disclosure limits context bloat and keeps performance and cost under control compared with native Copilot alternatives.
Practical takeaway: For SharePoint admins and builders, any Claude Skill can become an installed SharePoint Skill by placing it in the agent assets library and pointing it at lists or libraries.
This enables reusable, enforceable behaviors and faster, more consistent site experiences.

Copilot Studio: SharePoint AI Approvals

HubSite - Editorial — Mon, 06 Jul 2026 18:06:03 +0000

Copilot Studio, SharePoint and Power Platform power AI contract review with risky clause detection, auto approvals audit

Copilot Studio Workflows + SharePoint: Use Copilot Studio Workflows to add AI decision steps that trigger when files land in a SharePoint library.
AI reads documents and starts approval flows automatically.
AI Agent nodes and configuration: Configure AI Agent nodes with clear instructions, choose the model, and request structured JSON output so the workflow can act on extracted fields.
This avoids complex parsing steps and keeps the flow deterministic.
Document review and triage: The AI extracts key fields (counterparty, amount) and flags risky clauses like unfavorable payment terms.
Risky files route to Legal while clean files can auto-approve.
Approval routing and reasoning: Use Start and Wait for Approval actions combined with AI-generated reasoning to decide when to escalate to humans.
Keep human oversight for high-risk or complex cases to meet compliance needs.
Audit trail and history: Log every decision to SharePoint and present a popup View History using List Formatting JSON.
This creates a searchable, readable audit trail for reviewers and auditors.
Benefits and adoption: Expect accelerated approval cycles, more consistent decisions, and better handling of unstructured data (scans, images).
The feature is available in preview so teams can test and refine rules before full rollout.

Microsoft AI: June 2026 New Features

HubSite - Editorial — Mon, 06 Jul 2026 17:06:03 +0000

Microsoft AI update June: Azure hosts Claude, Anthropic in Foundry, Copilot and GitHub Copilot upgrades, Foundry IQ, Scout

Agentic computing: Microsoft frames a major shift from app-centric work to agents that execute multi-step tasks and workflows on behalf of users.
These agents act proactively, reason over long contexts, and reduce routine manual work to free people for higher-value activities.
Microsoft Scout: A new desktop personal agent that gathers emails, messages, and relevant files to propose actions and summaries.
Scout operates with user approval and ties into Microsoft 365 memory to deliver context-aware assistance.
MAI-Thinking-1 and model lineup: Microsoft introduced a new reasoning model, MAI-Thinking-1, focused on long-context reasoning and complex instruction handling.
The update also highlights Anthropic models on Azure, Claude Sonnet 5, and Kimi K2.7 in Copilot, expanding the model ecosystem for diverse workloads.
Microsoft IQ (Work IQ, Web IQ, Foundry IQ): A context and grounding layer that links world knowledge with enterprise data to make agents more accurate and relevant.
Work IQ and Web IQ provide APIs for developers; Foundry adds agent tooling, evaluations, tracing, memory, and new security/license controls for production use.
Project Solara & MDEP: Microsoft announced a new device category and the Microsoft Developer Edge Platform (MDEP) built to run agentic workloads with low latency and efficient resource use.
This hardware focus aims to improve real-time agent performance on dedicated devices.
Developer and product tools: Copilot Studio now includes improved agent evaluation, a toolbox, memory features, and orchestration updates for safer deployments.
Microsoft also highlighted Copilot Cowork, GitHub Copilot desktop and SDK, sandboxes, Frontier tuning, and other tools to help developers build, test, and govern agents.

Copilot Studio Drops Child Agents

HubSite - Editorial — Mon, 06 Jul 2026 16:06:03 +0000

Copilot Studio drops child agents expert tips to migrate and leverage Microsoft Copilot Power Platform SharePoint AI

Video overview: The video explains that the new Copilot Studio no longer offers Child Agents and replaces the old structure with a more unified Multi-Agent Orchestration approach.
It frames this as a strategic change to make agent interactions clearer, faster, and easier to maintain.
Legacy model limits: The previous Parent-Child Model routed user requests from a master agent to subordinate child agents, which often caused extra latency and unclear answers.
Delegation added complexity and made testing and updates fragile.
Key benefits of the new approach: Treating agents as connected peers improves Accuracy, reduces latency for faster Performance, and increases overall Reliability when you update or test individual agents.
The orchestration runtime now invokes specialized agents dynamically, reducing translation errors from intermediary routing.
Developer impact: Teams should stop planning around nested child agents and instead design agents as independent, callable components for easier Migration and safer Testing.
Expect to adjust routing logic and adapt any automation that assumed a strict parent-child relationship.
Immediate actions: Inventory existing child agents, document each agent’s responsibilities, and map how those functions become standalone peers in the new model (Inventory and Documentation).
Build and test small, independent agents first to verify orchestration and response quality.
Long-term strategy: Design agents as Interoperable Agents that share context via explicit Context Sharing patterns and monitoring hooks.
Monitor response quality and latency, keep domain logic modular, and plan incremental rollouts to reduce disruption.

Copilot Cowork: Set Spending Limits

HubSite - Editorial — Mon, 06 Jul 2026 15:06:03 +0000

Microsoft Copilot Cowork cost controls for IT leaders enable alerts, monitor use, protect budgets with SharePoint

Copilot Cowork is a metered AI service that runs complex, multi-step jobs and consumes Copilot Credits based on the chosen AI model, context size, task complexity, and whether the run is scheduled or manual.
Plan for usage-based billing instead of assuming these tasks are covered by standard Copilot licenses.
New tenants have Cowork turned off by default; administrators must enable the feature and configure billing and access before users can run metered tasks.
Assign admin roles and verify technical requirements before rollout to avoid unexpected access.
Use the Microsoft 365 Cost Management dashboard to set hard spending limits at tenant, group, and user levels so teams can’t exceed budgets without approval.
Define default budgets and tailor higher caps for teams that need heavier usage.
Enable customizable usage alerts to notify admins when spending hits set thresholds (for example 50%, 80%, and 100%).
Alerts help you act early — pause runs, increase limits, or adjust policies before costs escalate.
Implement a credit request and approval workflow so users can ask for extra credits when needed and admins can authorize critical overages selectively.
This maintains productivity while keeping spending under control.
Monitor consumption regularly and apply practical controls: pick appropriate models, limit context size, schedule heavy runs off-peak, and review budgets monthly.
These simple steps reduce surprises and keep Copilot Cowork adoption sustainable.

Copilot Studio: Create PowerPoint Fast

HubSite - Editorial — Mon, 06 Jul 2026 15:06:03 +0000

Copilot Studio for PowerPoint: Microsoft expert demo on building AI skills to auto-generate on brand decks from assets

Copilot Studio agents can turn a simple prompt into a polished PowerPoint (.pptx) file and return it in chat, with the final file saved to OneDrive or SharePoint for easy sharing and collaboration.
They generate full slides from prompt to downloadable deck without opening PowerPoint.
Key flow: the agent captures input variables (topic, slide count, tone, audience), uses a Generative Answer node to draft content, then calls a Power Automate flow that fills a PowerPoint template and creates the .pptx file.
This links the agent’s output to a reusable template and saves the result to cloud storage.
Main benefits include massive time savings, consistent corporate branding across decks, and automation at scale so non-technical users can produce high-quality presentations quickly.
Agents can tailor tone and slide count to match different audiences.
Templates and assets: flows retrieve a template from cloud storage, and an AI Builder prompt populates text boxes, bullet lists, and image placeholders with generated content and images.
Mapping input fields to template placeholders ensures consistent layout and design.
What’s new: unlike Copilot inside the PowerPoint app, Copilot Studio agents run as standalone bots that can be embedded in Teams, websites, or custom apps, letting users generate decks from anywhere.
This expands where and how teams can request presentations.
Best practices: define clear variables (topic, slides, audience, tone), keep branded templates updated, and verify outputs using Test Copilot before wide deployment.
Regularly review generated decks to ensure accuracy and brand compliance.

Azure AI: Hidden Supply-Chain Attacks

HubSite - Editorial — Sun, 05 Jul 2026 18:12:18 +0000

Protect AI agents with Microsoft Entra Agent ID and Defender to govern identity and block leaked API keys

AI agent ecosystem: Attackers target the invisible parts of agent platforms — agents, MCP servers, skills, packages, API keys, prompts, and identities — not just models themselves.
Watch for threats that hide in integrations and third-party components.
Attack vectors: Common weak spots include malicious packages, untrusted MCP servers, leaked API keys, prompt injection, insecure AI-generated code, and context flooding.
These vectors let attackers influence agent behavior without breaking classic defenses.
Agent ID: Use non-human or workload identity (for example, Microsoft Entra Agent ID) to identify and govern agents.
Identity helps but does not replace controls that limit what agents can discover or call.
Visibility & logging: Log agent actions, capture context, and enable replay to understand what an agent did and why.
Monitoring agents like threat actors helps detect fast, automated behaviors humans may miss.
Practical defenses: Enforce sponsorship and governance, vet packages and MCP endpoints, rotate and scope API keys, separate dev “vibe” environments from production, and require human review for risky actions.
Combine identity, least privilege, and runtime controls to reduce risk.
Attacker advantages: Threat actors gain persistence, invisibility, scale, and trusted access by exploiting agent ecosystems; agents can act faster than humans can review.
Prioritize detection and governance to counter these benefits.

Fable 5: Worth Double in Copilot?

HubSite - Editorial — Sun, 05 Jul 2026 17:12:18 +0000

Copilot Cowork: Fable vs Sonnet on SharePoint - dash, deck, Word, email; model choice spikes Microsoft Copilot costs

Copilot Cowork test: A YouTube demo ran the same prompt and SharePoint library through two models to produce an interactive HTML dashboard, a PowerPoint deck, a Word summary, and a drafted email from real ticketing data.
Clear cost gap: Sonnet 5 used 912.2 credits while Fable 5 used 1,738.4 credits, creating an 826-credit gap for the exact same job.
Real output differences: Fable 5 added a sticky header and more color on the dashboard; otherwise the results looked similar and the extra polish may not justify the higher cost.
Why it matters: Each Cowork task automatically picks a model and every model runs at a different rate, so model billing becomes a hidden line item on your bill.
Practical takeaway: Audit which model runs your workflows, prefer cheaper models for routine reports, and reserve costlier models for tasks that need obvious visual or creative gains.
Follow-up: The next newsletter will recommend which models to choose and when to switch to control credits while keeping output quality high.

Excel: Turn Tables Into Live Websites

HubSite - Editorial — Sun, 05 Jul 2026 16:12:18 +0000

Microsoft expert: turn Excel into a live no-code website with Power BI dashboards, drill-downs and live Excel connector

no-code: The video demonstrates turning an Excel table into a live, interactive website in minutes without writing code.
It shows simple tools that let anyone share spreadsheet data as a web app viewable in a browser.
Export to HTML: First step is converting the Excel table to a standalone HTML file so the data becomes a web page rather than a downloadable spreadsheet.
That export preserves table structure and enables sorting and filtering when supported by the tool.
Netlify: For deployment the tutorial suggests dragging the exported file to a free hosting service to create a public URL instantly.
Renaming the file to index.html and uploading it makes the site reachable without user logins.
Drill-down: The workflow adds interactive features like a searchable directory, dashboard views, and pivot-style drill-downs so viewers can click visuals and see underlying rows.
It also includes an Export to CSV button that downloads only the filtered results.
Live connection: The best practice shown is connecting the site to the source Excel file so updates refresh the web app automatically or with a quick re-export.
This keeps published data current and removes the need to resend static files.
Best practices: Prepare a clean table with a header row, convert ranges to an Excel table, and test filters before export.
Apply branding (logo, colors) and verify that filtered exports and drill-downs work before sharing the public link.

PowerPoint Morph: Easy Animated Diagrams

HubSite - Editorial — Sat, 04 Jul 2026 12:26:35 +0000

Microsoft expert PowerPoint Morph guide for animated process diagrams in medical workflows using Microsoft Office

PowerPoint Morph — The video demonstrates how to turn a static slide into a smooth, animated process diagram by duplicating a slide and letting Morph interpolate object movement and size changes.
Build the Base Diagram — Start with a clean layout: draw a horizontal line, place circular step markers, add numbers and short subtitles so each step reads clearly on the slide.
Duplicate Slide — Duplicate the starting slide, then move or resize the shapes on the copy to show the next state; Morph creates the motion path automatically between the two slides.
Animate with Morph — Apply the Morph transition to the second slide to animate steps sequentially; preview and adjust transition duration for smooth pacing.
Name objects — Give shapes distinct names or ungroup elements when you need independent motion; consistent naming helps Morph match and animate the right items.
Benefits and uses — Morph saves time, produces polished visuals, and works well for medical workflows, business processes, training, and teaching where clear step-by-step animation keeps the audience focused.

Excel: Dynamic Contents Page - No Macros

HubSite - Editorial — Sat, 04 Jul 2026 07:06:04 +0000

Microsoft expert: create a dynamic Excel contents page without macros with TOCOL, hyperlinks and Power BI via SharePoint

Dynamic table of contents
Build a navigation sheet that lists every workbook sheet automatically and creates clickable links to each page.
This gives fast access in large workbooks and updates when you add or remove sheets.
3D references
Use a 3D range like SheetStart:SheetEnd to include all sheets between two markers in one formula.
If you insert a new sheet outside that marked span you may need a small refresh, but sheets added inside update automatically.
VSTACK
Stack multiple arrays vertically to create a single column of sheet references or names.
VSTACK helps combine sheet name results from different formula parts into one dynamic list.
TOCOL
Convert a multi-row or multi-column spill range into a single column list for clean display and further processing.
Use TOCOL to turn VSTACK output into a neat, single-column table of sheet names.
TEXTAFTER and HYPERLINK
Use TEXTAFTER to strip the sheet name from a full reference and clean the display text, then wrap it with HYPERLINK to make each name clickable.
This pair creates readable labels and instant navigation without any VBA.
No macros — advantages & requirements
This formula-based approach avoids VBA security warnings, works in cloud/shared files, and stays portable across users.
It needs a modern Microsoft 365 build (Excel 2208 or later) and a simple setup: set start/end sheets, create the sheet-name formula, build the list, then add hyperlinks and helper columns for layout.

Custom Engine Agents: Vision AI and Policy

HubSite - Editorial — Sat, 04 Jul 2026 06:06:04 +0000

Custom Zava agent adds vision analysis and policy search with Azure AI Search, Azure Blob Storage and Copilot

Custom Engine Agents — The demo, presented by Ayca Bas, shows specialized Copilot agents that combine text, images, and structured data to automate complex tasks like insurance claims.
Vision Analysis — Agents use multimodal AI (including Mistral alongside GPT) to analyze photos and documents, extract key details, and generate short reports that speed up manual reviews.
Policy Search — Agents can look up and verify active insurance policies by number or owner, retrieve coverage details, and cross-check claims in real time to reduce processing delays.
Azure AI Search — A unified knowledge base indexes claims, policies, and blob storage content so the agent can answer queries that require combining text, images, and structured records.
Developer control — Teams can choose their own models, orchestrators, and plugins, integrate external tools, and enforce compliance while tailoring agent behavior to business needs.
Enterprise integration — These capabilities support scalable deployment across Microsoft 365 apps, improve decision speed and accuracy, and include monitoring and compliance features for production use.

Power BI: Watch AI Build a Full Report

HubSite - Editorial — Sat, 04 Jul 2026 00:25:50 +0000

Microsoft expert: Watch AI build a Power BI report and drive transformation with Power BI and Fabric training

Copilot in Power BI: This YouTube video demo shows how Microsoft’s Copilot builds full Power BI reports from simple natural-language prompts, and this summary explains the main features and steps shown.
Generative AI: The tool uses generative AI to create complete report pages, pick suitable visuals, and write short explanations, turning spoken or typed requests into ready dashboards in seconds.
Semantic model: Copilot reads the report’s underlying semantic model (your data and relationships) to decide which charts and calculations best show the insights, so visuals match the data shape and intent.
Time savings: The demo highlights big time savings—tasks that once took minutes or hours can often finish in under a minute, reducing the need for deep DAX or manual layout work.
Flexible editing: Users can ask Copilot to add, change, or remove visuals on an existing page, and built-in undo/redo and auto-generated narrative summaries help keep control and clarity.
Edit view: To use the feature in the Power BI service, open a report in Edit view and click the Copilot icon; you can then let Copilot suggest content or type a direct prompt for the visuals you want.

Azure Update 26th June 2026

HubSite - Editorial — Fri, 03 Jul 2026 21:34:37 +0000

Microsoft Azure update: VM application consistent restores, blob integrity, PowerShell Postgres, GitHub Copilot, Foundry

Confidential Live Migration for Intel TDX: Azure can now move confidential VMs to updated hardware with minimal interruption, keeping hardware-based confidentiality protections intact.
Azure Local small form factor (Public Preview): Microsoft added a compact deployment option for Azure Local to simplify edge installs in space-constrained sites like retail or transport hubs.
Foundry Local with vLLM support: Edge AI gains multi-node deployments and vLLM-optimized models, enabling stronger local inference without continuous cloud connectivity.
Azure Premium SSD v2 non-zonal support: Premium SSD v2 disks now work with non-zonal, single-instance VMs in selected regions, easing deployment choices for high-performance storage.
Azure Migrate discovers SMB and NFS file shares: The tool now assesses file shares on Windows and Linux, giving teams a data-driven view to plan migrations to Azure Files.
File share-centric management (GA): Azure Files introduced a resource-first model via Microsoft.FileShares, simplifying large-scale file share operations and management.

Microsoft Entra: Fix MSP Admin Access

HubSite - Editorial — Fri, 03 Jul 2026 21:34:37 +0000

MSP: end shared Global Admins and MFA vault use, secure Microsoft three sixty five with GDAP, Windows LAPS and Intune

Shared Global Admin and shared credentials destroy accountability and increase risk.
When multiple people use the same admin account or store MFA codes in a shared vault, you cannot trace actions or revoke access per person.
Permanent Global AdminGive staff only role-appropriate rights and avoid always-on tenant-wide admin roles to limit damage if an account is compromised.
Use Named admin identities, GDAP for partner access, and Privileged Identity Management (PIM) for Just‑in‑Time elevation.
Require a ticket or justification, set short durations, and log every elevation for auditability.
Protect admin sign-in and workstations with phishing-resistant methods like FIDO2, and isolate admin tasks on Privileged Access Workstations (PAWs).
Apply Windows LAPS so Global Admin credentials never live on client devices.
Tightly control emergency access with locked-down break-glass accounts and limit the number of active Global Admins.
Use approval workflows, MFA that isn’t reusable, and quarterly access reviews to keep roles current.
Audit these items this week: look for shared accounts, check active vs eligible admin assignments, remove legacy auth protocols, confirm GDAP use, and enforce phishing-resistant MFA.
Fixing these areas removes easy attack paths and improves compliance and traceability.

Microsoft Fabric: Data Integration and ETL

HubSite - Editorial — Fri, 03 Jul 2026 19:04:04 +0000

Microsoft Fabric guide to data integration and migration from ADF Synapse SSIS to Fabric Data Factory AI ETL future

Microsoft Fabric: A unified analytics platform that combines data ingestion, transformation, orchestration, and BI in one workspace.
It reduces tool sprawl and lets teams land data, transform with Dataflows or Notebooks, and analyze without switching systems.
Migration Path: Fabric supports mounting and conversion for Azure Data Factory and tools like Synapse and SSIS to minimize rework.
Organizations can move pipelines incrementally and keep existing logic while modernizing to Fabric Data Factory.
Data Integration Components: Key pieces are Dataflows (low-code transforms), Pipelines (orchestration), Copy Job and Mirroring (fast ingestion and replication), plus support for Apache Airflow and DBT where needed.
Use each tool for the right scenario: Dataflows for transform, Pipelines or Airflow for complex orchestration, and Copy Job for bulk loads.
SCD Type 2 & MCP Servers: Fabric simplifies common patterns such as SCD Type 2 via Copy Job features.
MCP Servers help run Data Factory, Power Query, and Dataflow workloads reliably and at scale.
AI Impact on ETL: AI and Copilot speed development with code suggestions, error checks, and migration assessments, reducing repetitive ETL work.
ETL roles will shift toward design, governance, and higher-value tasks that use AI as an assistant.
When and How to Migrate: Start migration when you need unified governance, faster delivery, or lower operational overhead.
Adopt an incremental approach: assess current assets, use automated assessments, migrate non-critical pipelines first, then move complex workloads after testing.

Fabric Apps: Build No-Code Data Apps

HubSite - Editorial — Fri, 03 Jul 2026 18:04:04 +0000

Microsoft Fabric Apps let devs build data apps in Fabric with Power BI, TypeScript, generated APIs, hosting and auth

Fabric Apps bring app development directly into Microsoft Fabric, letting teams build data-driven applications without wiring separate backend services.
They combine data models, APIs, authentication, and hosting inside the Fabric workspace.
Developers define data models in TypeScript and use the Rayfin SDK to generate backend pieces and client libraries automatically.
This flow creates database schemas and GraphQL endpoints so you spend less time on boilerplate.
Fabric Apps eliminate much of the traditional backend setup, enabling rapid prototyping and faster time to a live URL.
The platform provides managed hosting and scaling so teams can focus on app logic and UX.
Apps run as first-class Fabric artifacts and inherit workspace governance and security controls with direct access to OneLake data.
That makes permissioning, compliance, and data access consistent with other Fabric items.
The development experience uses type-safe clients and TypeScript decorators to catch errors early and reduce runtime bugs.
Fabric Apps also suit AI agents and stateful data services that need persistent storage and structured APIs.
Use Fabric Apps when you want interactive data experiences beyond dashboards—apps that read and write Fabric data and ship quickly.
Note: the feature is in preview, so review the docs and try a small prototype to evaluate fit for your workflows.

Excel AGGREGATE: Its Not What You Think

HubSite - Editorial — Fri, 03 Jul 2026 12:23:39 +0000

Microsoft Excel expert guide: why AGGREGATE returns errors, arrays versus ranges and workarounds using BYROW and LAMBDA

AGGREGATE: A single Excel function that replaces many separate formulas by offering 19 aggregate operations (average, sum, median, percentiles, quartiles, etc.).
It debuted in Excel 2010 and acts like a more flexible, powerful version of SUBTOTAL.
Syntax: =AGGREGATE(function_num, options, ref1, [ref2]).
Use the reference form for normal ranges and the array form in modern Excel (Excel 365) to accept expressions like A1:A10*(A1:A10>0).
Key features: Options (0–7) let AGGREGATE ignore error values, hidden rows, or nested functions so you still get a result when parts of the data contain errors or are filtered.
Example: option 6 tells AGGREGATE to skip error values when calculating a sum or average.
Limitations: AGGREGATE expects vertical (column) ranges for its primary reference and does not let you ignore specific row numbers. Horizontal ranges or incompatible arrays often cause #VALUE! or unexpected results.
Be careful with whole-column references or mixed array/range inputs, which can trigger errors.
Basic workaround: For all Excel versions, clean the data first by using helper columns or IFERROR to replace error cells with blanks or valid values, then run AGGREGATE on that cleaned range.
Using a helper column to normalize horizontal data into a single column also avoids many common failures.
Advanced workaround: In modern Excel, reshape or process arrays with functions like BYROW and LAMBDA, or convert ranges to a single column with TOCOL/TOCOL, then apply AGGREGATE’s array form.
These techniques let you handle horizontal data, complex filters, and row-by-row logic without helper columns and without Ctrl+Shift+Enter.