AI Vendor Management - Formiti

AI is going crazy, and you can build your own but generally you need to look at a supplier, so it's worth understand management of Vendors, you as the controller using their service are at risk of them not making their AI operations transparent.  It's a big business risk to my clients.  

GDPR is closely linked to AI, and if you use a service/vendor, the reputation and fine risk may fall on you as the provider.  Need visibility into each vendor, how they are using AI, in turn they are using vendors so it's a nice complex dependency problem.  You need to be aware of what you are relying on.

Ensure contracts with vendors consider AI, how the process your data and how their sub process vendors do the same.

Track website customer behaviour, we use a vendor to clean up the data.  In turn, I have no idea that they are using AI outside of the UK or EU.  Follow the dependency chains as all this needs to be transparent to the end customer if needed.

Microsoft's Copilots Explained as the naming kills me

Microsoft's Copilots products... starting with M365 Copilot options

Source: Screenshot from a presentation, 6 May 2025 - must come from MS.

1. Microsoft 365 Copilot versions

1.1. Copilot (Microsoft 365 Copilot Chat) (Is Free)

It allows access to ChatGPT-4 with web content.  AI like ChatGPT, Perplexity, ...

1.2. Microsoft 365 Copilot Pro

(+$20/month full retail if bought as an add-on).  More features but not everything the M365 Copilot for Business offers.  All users get Copilot Builder (maker capability), and consumers need the M365 Pro licence to use M365 Copilot agents.

1.3. Microsoft 365 Copilot for Business

I believe a lot of documentation and people are grouping "M365 Copilot Pro" and "M365 Copilot for Business" together, and only see a free and pro/paid versions of M365 Copilot. 

M365 Copilot Pro for Business (I know :) ) is grounded in the business data you can access in your enterprise.  Inline copilots, including grounded business data inside the office applications.

Microsoft 365 Copilot is an AI-powered Agent with multiple child Copilots for Microsoft 365 apps like Word, PowerPoint, Outlook, Excel, and Teams. It leverages large language models (GPT-4 and GPT-4Pro) and your enterprise data through the Microsoft Graph. To work with M365 Copilot Pro, you need the M365 Copilot license for each user accessing or creating the new agent, which is pricy, per user per month. 

Tip: Optimise results by preparing content!  Like docx, pptx and pdf files. PDF, DOCX, XLSX (kind of), PPTX. Also works on CSV, text, HTML, MD, and WAV audio files.  Supported file type.

Microsoft-specific app copilots included in M365 Copilot Pro:

• 2.1. Word Copilot: Helps with drafting, rewriting, editing, summarising, and generating new ideas.
• 2.2. Excel Copilot: Assists with data analysis, generating insights, creating complex formulas, and automating tasks.
• 2.3. PowerPoint Copilot: Aids in creating presentations, suggesting layouts, creating slides, and enhancing visual content.
• 2.4. Outlook Copilot: Supports email management, drafting responses, scheduling, and organising tasks.
• 2.5. OneNote Copilot.

Note: Microsoft Copilot does not allow uploading images/pictures like ChatGPT does as of March 2025, but I don't doubt it is coming.  ChatGPT can also redraw (using DALLe)

Extending the use of M365 Copilot Pro licensing

1. M365 Copilot takes a no-code approach to building agents with our company data. Each user needs an M365 Copilot Pro licence to access the new bot/agent/copilot.
2. Copilot Studio is a more advanced way to build copilots and requires a separate license.

Options Building you own Agents/Copilots with business aware data.

Agent Builder - all users need the M365 Copilot Pro licence.

Copilot Studio - makes need a Studio Builder licence for each maker, build agents using low code AI.  Each query is charges using messages as it's currency.  There is complex pricing for instance a query can cost 13 message credits.

AI Foundry - the individual paid services on Azure relating to AI uses tokens for billing.  A simple query is 4 tokens depending how the question is broken down.  Once again it is difficult to estimate the costs.  

2. Microsoft Dynamics 365 Copilots

Overview of Dynamics 365 Copilots:

Source: Microsoft (I lost the reference)

2.1. Customer Engagement (CRM) Apps

1. Dynamics 365 Sales – Manage leads, opportunities, and customer relationships.
2. Dynamics 365 Customer Service – Case management, knowledge base, and omnichannel support.
3. Dynamics 365 Field Service – Manage field operations, work orders, and technician scheduling.
4. Dynamics 365 Marketing (now part of Customer Insights - Journeys) – Campaign automation and customer journeys.
5. Dynamics 365 Customer Insights:
   • Data – Unify and analyze customer data.
   • Journeys – Design and automate personalized customer experiences.
6. Dynamics 365 Customer Voice – Collect and analyze customer feedback.

2.2. Finance & Operations (ERP) Apps

1. Dynamics 365 Finance – Core financials, budgeting, and global accounting.
2. Dynamics 365 Supply Chain Management – Inventory, manufacturing, and logistics.
3. Dynamics 365 Project Operations – Project planning, resource management, and billing.
4. Dynamics 365 Commerce – Unified retail, e-commerce, and POS.
5. Dynamics 365 Human Resources (being merged into Finance) – HR management and employee self-service.

2.3. SMB-Focused App

• Dynamics 365 Business Central – All-in-one ERP for small and medium-sized businesses (finance, sales, purchasing, inventory, and more).

2.4. Industry-Specific Solutions

• Microsoft Cloud for Industry (e.g., Healthcare, Financial Services, Manufacturing) – Built on Dynamics 365 and Power Platform with tailored capabilities.

Note: Microsoft D365 Copilot/ Microsoft Copilot for Dynamics 365 and Power Platform/ Dynamics 365 AI are in this area.

3. Copilot Studio

Build copilots and distribute them to the business. This is awesome, but from a naming perspective, it actually makes sense. 

4. Azure AI Foundry

Organisations and developers can use the AI Foundry platform to build AI-driven solutions.  Can use 1,800 models and 200 Azure Services.  There are so many AI services on Azure.  I'm not even starting that conversation.

5. Security & Governance of Microsoft 365 Copilot 

The M365 copilot will respect the Access Controls you have permission to, so Jim in operations won't see Tim's sales figures in a Word document in SharePoint. Jim does not have access to the document from the sales site collection. Therefore, they can't query Copilot for the data they don't have permission to (done using MS Graph).

Purview: DLP improves what your users can do, and better governance.  Ensure M365 copilots are used ethically.

Understand how users are using your M365 Copilot.

6. Code Copilots

GitHub Copilot

GitHub Copilot is an AI-powered tool that helps developers write code by suggesting snippets and completing lines in real time. Developed by GitHub and OpenAI, it learns from public code repositories and supports various programming languages and frameworks. Its goal is to boost productivity and reduce repetitive coding tasks.  It's amazing!! and getting better all the time.

Code Copilots: GitHub Copilot extensions in VS and VS Code. Amazon Q Developer is an AWS code generator that is best for building code and solutions. GitHub comes in three flavours: Individual, Business, and Enterprise.

GitHub explains code, optimises your code's performance, improves readability, generates unit tests, improves error handling, adds new code with requested functionality, ensures coding consistency, and improves modularisation to help with the DRY principle.  GitHub is the diggity bomb.

Note: ChatGPR is owned by OpenAI.  Microsoft is a significant investor but does not own the service.

7. Summary

Naming and complexity with Microsoft are tough. The AI parts are massive. The M365 copilots break down into Dynamics or O365 worlds, each with many options.  I wish Microsoft had a hierarchical, sensible naming system, which no doubt the Marketing department would hate.

AI Copilot comparrison

 Lunchtime play ...  Claude.io was the best I used... Deepseek didn't work....

AI Engine Comparison Test:

I recently asked a few AI engines a complex tax query:

Perplexity Pro (paid): While I love the app, the answer was not great 6/10

ChatGPT (free): Excellent, factually correct, not too clear in one area 8/10 

DeepSeek: I registered and tried the search, but got the result "The server is busy. Please try again later."  I tried a few times, it is working for common queries, but I assume as this logic wouldn't be cached, it can't even have an attempt with the free version 2/10.

Bing/M365 Copilot: Got a fairly similar result to ChatGPT missing an option and not as well laid out 7/10 

Claud.io: Clean result that is factually correct and offered possible items that would affect the calculation 10/10

Perplexity is my default AI power Search Engine:

Perplexity and the others are great for teaching me this (I use a homemade flip learning type approach), and here are my "follow-on" questions. However, to truly understand something, I still use books (mixed with perplexity), but nothing beats a human-written or recorded topic to give the best understanding.

Prompt Engineering with Microsoft Copilot

Prompting Microsoft Copilot: Prompting is asking the copilot to give you information.

Four parts to cover in your prompt> 1. goal 2. context 3. source 4. expectation

1. Goal - what do I want, e.g., tell copilot what we want

I want tactics for pricing my SaaS products

2. Context - Why do I need it

e.g., this is a new product in the B2C space

3. Sources - optional - ask from specific resources

e.g. provide my company stats and ask to compare to my market (from the LLM)

4. Expectations - Table of info, summary, or even an image(maybe not)

M365 Copilot app.

Apps that M365 Copilot embedded with if you have the M365 licence include: 

1. Word, 2. Excel, 3. PowerPoint, 4. Outlook, and 5. Teams

1. Word

Word Copilot is part of the M365 copilot branding.  Generates scaffolding/first draft.

2. Excel

Use in Excel with your own data - allows easy query using English/natural language.

3. PowerPoint

Draft a presentation using org themes based on prompts. You can also use Copilot to add images you create using Copilot (DALL-e).

My First Mendix Widget - Azure AI component to convert Text to audio

 https://www.youtube.com/watch?v=0bxJeYSDV28 (3min video)

Mendix Marketplace - Azure AI Services - Convert Text to Audio  - Here is is in the Marketplace.  Thanks to Ameer for helping me build the widget.

Basics to setting up an AI Transformation Program - High level

Overview: Many large organisations have started on their Artificial Intelligence (AI) journeys, many have started in weird directions and some are taking the wait and see approach.  As a general rule, I feel most organisations should try identify the most important use cases for AI and using a basic scoring system take the easy high values use cases first. 

A good option is to get a AI focused team to put the AI program into your business.  Get people seconded as they have key domain knowledge and couple with IT experts pref with strong AI skills.  

1. AI Idea Generation: Collate a detailed list of possible ideas.  I like to use SharePoint, it's a good idea to open ideas up to the business or make gamification of idea generation to get a good set of ideas.  You;ll get lots of overlap, distill into unique ideas, and get the team/stakeholders together.

2. Business Impact: how does this help our customers or generate revenue or does this improve our operations.  Can this be quantified?

3. Technological Feasibility: Can technology meet the requirement, and at what cost?  High level technical design is strongly advised as show below in figure 2.

4. Implement: If the idea holds up to Business impact, and is possible, select the highest value to lowest effort ideas.  It is a good idea to start with the simpler ideas to get the ball rolling.

Figure 2. HLD example

Note:  This post is looking at building AI to improve your business, AI also affects all internal staff, so train them.  Allow them to use AI tools like ChatGPT, and train them not to expose internal propitiatory information.  Give your users the correct tools to be the most effective.

Ignite 2023 - Keynotes - Summary

15 & 16 November 2023

Good Keynote: AI is driving a lot of innovation.

Microsoft Fabric in GA (25k instances already).  New feature is 'Mirroring' - copy cosmos/SQL et al into Fabric.  OneLake.  Can bring lots of data from multiple sources into Fabric in near rela time.

MS Teams (320 million users):  Bring everything to the user in one place, not just communications but a canvas for apps.  Good place to build line of business applications.  New teams app - way faster, easier to use.  Teams Premium - intelligent meeting recap is working well, can integrate recap with copilot. 

Copilots - needed for nearly everything you do.  understand context of where you are.  MS has hundreds of copilots.

Copilot Studio - Custom GPT's, can add plugin's to add your own data, can hook into an enterprises unique data.

Copilot for Service - allows agents to get information to provide support, looks interesting.

Personal thoughts: AI is going to be a mega trend that will influence the world hugely, there will be lots of weird decisions on the journey. Currently, it is mainly proving useful as another tool to help improve existing processes.  AI helps me work faster and spend my time on exploration rather than bring base understanding together.

Part 2 Keynote:

Microsoft Graph gives the copilot context within an organisation.  Use plugins to add enterprise data or Open AI GPT's.

Surface Pro Hub 3 released  - looks good, rest of the hardware looks higher spec.

SharePoint Premium - improved knowledge and content management on SharePoint.

Copilot Studio - useful to build internal copilots.  1. Connect copilot to other systems using plugins or GPTs 2. create workflows 3.  Controlled by IT.

Copilot Studio overview

Mesh - Teams can join immersive experience/events, not sure what this means.  GA expected Jan 2024.

Microsoft 365 copilot release to GA 1 November. 

Why Copilot? MS are describing it as a productivity multiplier.  Allows users to be more productive and more creative.  Improves quality of work, avoids searching - as expected.  Makes mistakes but is getting better.

Microsoft Copilot - Bing chat is just MS copilot

So when logging use Entra Id (Azure AD), get contextual enterprise information.  Inherits security and privacy policies.  ACL controlled.  Includes MS graph and Apps.

Try it out: Copilot.microsoft.com  - Chat data is not saved/stored by MS.  Change from "web" to "work".  Also available in Windows taskbar

Ability to use copilot to pull in information, show more graphs, get data.  Good word example getting data from a pptx.

Great example of querying Excel using copilot, created a pivot table. Contrived but looks good, added rainfall from web to look at sales.  Powerful.

Never thought of copilots for being a participant in a meeting - might us amazing.  In teams meeting, takes real time notes, and pulls in info and summaries points for next meeting.  Add as a collaborative partner, can visualize discussions on a meeting whiteboard.

Loop - flexible collaboration, now with copilot.  Not my area but sounds impressive - i don't get it.  People, working with people, now also working with copilot.  Okay.

Copilot for Sales - looks promising.  Hooks into existing CRMs.

Copilot for Service - working with customers, get data that is correct to solve customers problems.  Concise summary, helps craft emails, updates CRM.  Looks very interesting!

Viva - Microsoft copilot dashboard powered by Viva - not sure on this topic.

Summary: People using copilot don't want to loose it.  AI is bringing big changes to many industries.  Promise is to take the grind out of work - sounds great let's see.  Copilot/AI will be a tool and shape how we work.

Microsoft Azure Artificial Intelligence (AI) - AZ-900 Notes

1. Artificial Intelligence (AI) 

• AI making PC behave like human intelligence.  
• Teach PC to do task for us.  
• PC predicts using patterns and can act.  And good at looking for anomalies.
• PC uses camera/photos to look for patterns.
• Engage in useful conversations, use multiple sources of knowledge.

2. Machine Learning (ML)

• Train PC's to see patterns and see patterns, and look for anomalies.
• Example. predict stock prices by looking at factors that affect stock price.
• Anomaly Detection - Detects unusual patterns e.g. CC used in Asia when normally in Europe, but transactions 10 min apart.  Therefore likely to be fraudulent.  Sort rubbish.
• Predictive models by finding relationships.  Give model data and train the model.  
• Example: flowers have features/characteristics e.g. colour, size, no petals, ...
• Using data to teach machine
• Supervise ML - need quality data including labels.  Avg humidity, hrs sunshine, rainfall, temp, month of year (features), ice creams sold(label/class), so we feed in temp is Regression ML.  Patient has features (weight, sex, age, bmi,...) give value btwn 0-1 of the person developing diabetes.  is Classification ML
• Unsupervised ML - data is not labelled.  Just feature provided, will group into clusters.  Pulls data out and figures out it's own criteria is Clustering ML.  Useful for fraud detection.
• Training - good data based on a training set and a validation set.  Train model, with most data, then check with remaining - allows us to see how close to what happened.  Service tries to figure out relationships.  Model is used by test data - see how close/useful model is.

3. Compute Vision 

• Self driving cars, sorting. Sort rubbish.
• Facial recognition, object recognition,..
• How do computers see?  picture is cut up into pixels, data is pulled and used to find possible ans. 
• Some types on Azure:  object detection i.e. car, bike, car, bus.  Image classification i.e horse, car.  Semantic segmentation i.e. Teams blur background.   Image analysis contect by bring various tougher.  Face detection.  OCR - read image and converts to text.

4. Natural Language Processing (NLP) 

•     interpret e.g grammerly, spam check, Alexia, 
•     Knowledge Mining - Extract info from knowledge and gain insights e.g. social media marketing.

Principals:

1. Fairness - ensure bias is excluded e.g based on gender.
2. Reliable & Safety - need high confidence and in certain systems cannot fail e.g. health systems, autonomous cars.
3. Privacy & Security - Ensure data is protected and not giving away sensitive data.
4. Inclusiveness - should be fair i.e. VI users
5. Transparency - what is the model based on, what could be an issue.
6. Accountability - who is liable for AI decisions

Azure:

1. Scalability & Reliability
2. AI Resources: sit in an Azure Resource Group

AI Services in Azure:

1. Azure Machine Learning - Developers to train, test and deploy ML models.  Within a subscription, create a Azure ML Workspace (consists of: compute, data, jobs, models) can then publish as a service.  Azure ML Designer, used for creating ML pipeline, data in to train model.  Automated Machine Learning user only needs to provide the data and select the model to use, service figures it out. 
2. Cognitive Services - vision, speak, language, decision.  Rest API endpoints - have already been trained, choose the model.  Can deploy multiple parts individually or together.

    3. Azure Bot Service - develop & managing intelligent bots like chat-bots
    4. Azure Cognitive Search - Data extraction, & enrichment for indexing.  Makes data searchable.

Anomaly detector resource - wizard to setup - Add Keys and endpoints to allow access.

Create a new Azure Machine Learning Service, will create a Workspace.  Use multiple azure services such as key-vault, AI, storage accounts.   

• Launch Studio
• Add Compute Cluster
• Add Data (csv, spreadsheets, nearly any form,...)
• AutomateML (figure it out without me) and run job
• Will show trends
• Deploy the model (i.e. to a web service)
• Shows "Endpoints" - get url and a test rig.

An brief introduction with two demos on OpenAI

OpenAI has a couple of service such as ChatGPT, and DALL-E.  The recording below, shows two a demos: 

• ChatGPT to gain insight, and
•  DALL-E to generate some artwork.

https://youtu.be/TdGjp171wAk - 3min 48 seconds

There are other suppliers of large scale AI engines such as Googles PaLM 2.

Updated Dec 2023: 

I created an Azure Open AI Service instance.  Very easy to access using the API's and nice to play with.  There is also the playground, and I generated 8 images of a train emerging from a tunnel in eight different artistic styles.  The hardest part of AI and DALL-E is framing the Prompt/Question.

OpenAI Studio on Azure AI using the DALL-E playground.

Dynatrace Product Play

 Dynatrace is pretty similar to Azure Monitor.

• Dynatrace (really good if you use multi cloud) Dynatrace - Saas offering is on AWS.  Can be on-prem.  
• Making workloads observable is using Logs, Traces, Events, and metrics into Dynatrace.  From these ingested events we analyse and can automate behavior. 
• OneAgent is deployed on the Compute i.e. VM, Kubernetes.  Can import logs from other SIEMs or Azure Monitor, so you can eventually get Azure service logs such as App Service or Service Bus.  
• Does Full stack and includes code-level and applications and infrastructure monitoring, also can show User monitoring.  
• Dynatrace offers scalable API's that are sitting on Kubernetes.  
• "Davis" is the AI engine used to help figure out the problems.  
• Alerting is solid.  
• Dynatrace can log against 1) network/Infra 2) SDK 3) DEM (User monitoring,..)  logs, traces, metrics are ingested either using OneAgent or OpenTelemetry.
• Management Zones - user only see's information they have access to and need.
• Define a Site Reliability Guardian (SRG) to each program/project, this allows you to identify thru RAG boards the current and recent state of the various pieces.  There are Guardian templates to use as a starting point.
• W3C Trace Context is used - it allows for end-to-end tracing.  OpenTelemetry or Dynatrace keep the trace and provide in headers (traceparent.
• Create documentation and tutorials for Dynatrace.  Dynatrace has a playground tenant for playing on.

High-level Architecture hosted on AWS.

High-level architecture for capturing logs et al. and then using the data.

Product Screen Shots:

Azure & Dynatrace

• Abnormality detection using AI. shall greatly improve observability and security. 
• End-to-end visibility is what makes it so amazing.
• Enterprises often use Dynatrace as there central SIEM solution, shipping from Azure in Dynamics takes planning but works well, categorise and ensure the right into is pushed into Dynatrace.  
• Dynatrace is the leader in Gartner and Forrester in it's space.
• Grail - lake house, schema-less, allows for easy fast query.  Massive scale.  Bring all data together and query at hyperscale.  Grail is in 15 regions either on AWS, Azure, or GCP for customers to use.  UK looks like AWS only. 
• Grail: Record level protection, masking data, support access controls (elevate privileges).  

Dynatrace architecture for Grail from Barcelona conference 5 Oct 2023.

Collect all events in Grail, automate the process of identify suspicious activity relating to security.  Faster reaction time.

Azure offers Dynatrace as a SaaS service

Updated 16 Feb 20224