Author: Magnus Oxenwaldt

Use ChatGPT to automatically update your team chat messages for clarity.

As we continue to work remotely in the wake of Covid, communication through text is more common than before. With messages flying back and forth, there is a chance of misinterpretation, which can cause serious issues. Messages can easily be misinterpreted, assigning incorrect intent or meaning. To minimize this risk , I created a Power Automate that checks my Teams messages for any risk of misinterpretation.

To achieve this, I used the newly released ChatGPT API. ChatGPT is a large language model trained by OpenAI to understand natural language and generate human-like responses. By leveraging ChatGPT’s capabilities, I created a Power Automate that analyzes the text of the messages and checks for any potential misunderstandings. If there was content in my chat message deemed to include ambiguous content, too many alternative meanings, or obvious spelling mistakes – then the Power Automate flow would update my Teams chat message.

To accomplish this, I used;

• A custom connector to OpenAI ChatGPT API
• A custom connector to Microsoft Graph API
• A Power Automate flow

The first step was to set up the Power Automate flow. I started by connecting it to my Microsoft Teams account and setting up a trigger that detects new messages. To not check all of them during proof-of-concept, I have added a condition that it will only check messages where I include a code. I put # start of the message as the trigger to check my text. Then, I added the action to get the text of the message. Next, I created an action to call the ChatGPT API through my custom connector. I passed in the text of the message as input and requested an analysis of the message and added the instruction for ChatGPT, so it was clear to know what to do with the text content. The API then returned a response with the analysis, including any potential risks of misinterpretation and a suggested rewrite. Finally, I added a condition to check if the analysis indicated any risks of misinterpretation. I let ChatGPT rate it between 1-100 and along the way I found that above 75 was a good rating threshold for me to have ChatGPT do rewrites. I also made sure to set the temperature parameter low to minimize the creativity in the responses – making them more predictable.

Sending the teams chat message to ChatGPT API with an instruction prompt.

Over the years, I’ve developed a lot of programming skills through various projects and assignments. However, as time passed, some of these skills became less useful or outdated, and I found myself struggling to find new and exciting ways to stay engaged with coding.

That’s when I discovered GitHub Copilot, an AI-powered code completion tool developed by OpenAI and GitHub. Using machine learning algorithms, Copilot is designed to help developers write code more efficiently, by suggesting lines of code and even entire functions based on the context of the code being written.

At first, I was skeptical about using an AI-powered tool to help me write code. I felt like it might take away from the creative aspect of coding, or make me feel like I wasn’t really doing the work. But after using Copilot for a few weeks, I quickly realized that it was a game-changer.

The AI models that power Copilot are incredibly accurate, and they seem to have an almost intuitive understanding of the code I’m trying to write. As I type, Copilot provides helpful suggestions and code snippets that save me time and frustration. It’s like having a programming partner who can anticipate my every move and help me write code faster and more efficiently.

But what I love most about using Copilot is how it’s reignited my joy for coding. It’s allowed me to focus on the parts of programming that I truly enjoy, such as problem-solving, rather than getting bogged down in the nitty-gritty details of syntax and formatting. With Copilot taking care of the grunt work, I can focus on the creative aspects of coding, such as designing algorithms and exploring new approaches.

Of course, Copilot is not a magic bullet that can solve all of my coding problems. There are times when it makes mistakes or suggests code that doesn’t quite fit what I’m trying to do. But overall, the benefits of using Copilot have been overwhelming , and it has boosted my confidence and helped me stay engaged with coding in new and exciting ways. Nowadays when I think of designs I often turn to AI and Copilot to quickly build that proof of concept required for me to feel confident about my designs.

In conclusion, if you’re a developer who’s struggling to find new ways to stay engaged with coding, I highly recommend giving GitHub Copilot a try. This tool has been a game-changer for me, it revamped my use for coding and I’m excited to see how it continues to evolve and help other developers find joy in their work.

Summary of below blogpost by Turbo –
The author of this text writes about their experience with ChatGPT and its premium version, Turbo. They discuss the differences between the various models available on Azure OpenAI and mention that they have found the speed of Turbo to be much faster than the standard version. The author also highlights how ChatGPT has become an indispensable tool for them in their daily life for tasks such as summarizing large texts, learning new things, and translating texts. They also recommend using the Chrome voice control extension for an even more convenient experience. In conclusion, the author encourages businesses to invest in OpenAI chatgpt services as they believe that AI will play a crucial role in the future.

ChatGPT has had issues with accessibility. No figure, given the mindblowing 100 millions users after 2 months.

But fear not, salvation is here! I acquired my premium account with Chatgpt for a measly 25 USD per month. This gives me access to faster response and more uptime (hopefully).

Now what is truly the difference? Well….let us ask ChatGPT Turbo 🙂

First attempt…ouch! Are there already 100 Million premium users?

My first attempt of asking Turbo was a failure. I think that this is intermitten given that Turbo is just out and I would think everyone is testing at once.

My second attempt was more fruitful.

This statement from Turbo I believe is simply not true?! But nevertheless we can see some differences in the url used to access turbo.

As we are aware, there are multiple models. Let us briefly have a look at which models we can easily access and the differences.

In your Azure OpenAI we have access to the following models when we want to build AI induced applications.

We can also get some more information either in OpenAI official documentation or in Microsoft Docs.

https://learn.microsoft.com/en-us/azure/cognitive-services/openai/concepts/models#gpt-3-models

Regardless of the models used in which version – I can already now say the speed is many folds in comparison to the standard, but I have no actual proof it is providing more nuanced responses. From what I see using the standard chat page – I can’t adjust any parameters. Which I have not as of yet needed. I simply ask ChatGPT to adjust its answer according to temperature and max tokens, where the temperature is creativity, and max tokens is the length of the response from chatgpt.

I have in addition to the standard page built my own apps to access the various OpenAI API’s but as of now my default homepage is https://chat.openai.com/chat, and this is where I go 9 times out of 10.

ChatGPT has become my number-one place for the following daily tasks;

• Summarize large texts.
  I simply don’t always have the time to read long texts. I use summaries to help me get information quickly or as a teaser to actually prioritize time to read the text in its entirety.
• Learn new things
  Getting responses and teaching me in a dialogue fashion instead of learning texts from standard documentation has helped me greatly. Oftentimes I learn new things by first trying to grasp the holistic essence of a concept before I spend more time drilling down into the details.
• Text translations
  I work with multiple languages daily and documenting and understanding text is important. My go-to language is English for documentation, but I do document in other languages every now and then, and the translation of longer texts with GPT3.5-based models has in my experience proven vastly superior over Google.

Another tip from me is to use the chrome voice control extension. It supports many languages. It allows me to talk to chatgpt without using my keyboard. This has proven very helpful and raised some eyebrows on the train to work!

My recommendation to all businesses out there who care to power up their colleagues – is to allow the cost of OpenAI chatgpt plus services to become deductible. It has been said before and let us say it again – AI will not replace you in the near future, but the people who use it will!

Happy ChatGPTying

Disclaimer: This blogpost was partially written by TURBO!

The recent Dynamics 365 FastTrack TechTalks topic was Power Platform Admin Center Integration with LCS. The integration has existed for some time, and development is being done. But the major takeaway is the future existence of admin tools within LCS to deploy and maintain Dynamics environments with the current One Dynamics Admin platform strategy. Not far into the session following statement was proclaimed.

Move everything so that we have one unified admin experience

In a nutshell what this means is that in 2-3 years we will see a change where no new Dynamics environments are deployed in LCS – but all deployment and maintenance is purely done from PPAC. Where Dynamics 365 Finance is an app in a shared powerplatform environment – rather than its own standalone environment.

One Dynamics – One Platform

If you wish to view the entire sessions, please check out below urls.

https://community.dynamics.com/365/dynamics-365-fasttrack/b/techtalks/posts/one-admin—lcs-ppac-integration-december-7-2022

Most Connectors win!

Microsoft announced recently that they surpassed 1000 connectors. This might not mean much to many, but successful digital transformation projects are oftentimes riddled with non-working integrations, especially during transitional architectures. To have access to vendor-supported cross-application integration connectors is oftentimes not only a selling point long-term but also what is required to meet the demand of time to market. No one around today wants to invest in their own bespoke integrations unless it is necessary.

Dynamics 365 and the rest of the Microsoft Application stack connect today to any known application. Either by an application-specific connector or via a more generic connector.

What is a connector?

There are heaps and heaps again of different types of connectors. But within Microsoft, we commonly refer to the Azure or power platform connectors.

The integration of Microsoft Dynamics 365 with connectors provides businesses with a powerful platform to automate their business processes, streamline their workflows, and enhance their overall productivity. The combination of Dynamics 365 and connectors allows businesses to seamlessly integrate their cloud-based and on-premise applications and enables data exchange between these systems in real time.

What are Azure Connectors?

Azure Connectors are a set of pre-built, managed, and configurable services that provide a fast and secure way to integrate applications and services. These connectors provide an easy-to-use interface for exchanging data between systems, and are designed to work seamlessly with Azure services and other cloud-based and on-premise applications. With Azure Connectors, businesses can eliminate the need for custom integration code and reduce the time required to integrate systems, which ultimately leads to increased efficiency and cost savings.

How do Azure Connectors work with Dynamics 365?

Azure Connectors provide a bridge between Dynamics 365 and other applications and services, enabling the exchange of data in real-time. The connectors can be used to integrate Dynamics 365 with a variety of applications, including SharePoint, OneDrive, Excel, and many more. With the use of Azure connectors, businesses can easily connect Dynamics 365 to their existing applications and services, allowing them to automate workflows, streamline processes, and enhance the overall productivity of their organization.

What are the benefits of using Azure Connectors with Dynamics 365?

1. Increased Efficiency: Integrating Dynamics 365 with Azure Connectors eliminates the need for manual data entry and reduces the time required to complete tasks. This increased efficiency enables businesses to focus on more important business activities, leading to enhanced productivity and cost savings.
2. Improved Data Quality: Azure Connectors provide a secure and reliable way to exchange data between systems, ensuring that data is accurate and up-to-date. This improved data quality provides businesses with a single source of truth, which can be used to make informed business decisions.
3. Enhanced Collaboration: Integrating Dynamics 365 with Azure Connectors enables businesses to collaborate more effectively by providing access to real-time data from multiple systems. This enhanced collaboration leads to improved teamwork and better decision-making.
4. Customizable Workflows: Azure Connectors allow businesses to automate workflows, streamline processes, and enhance the overall efficiency of their organization. The connectors can be configured to meet the specific needs of each business, providing a tailored solution optimized for their unique requirements.
5. Enterprise Support: Whenever you need a supporting hand – Microsoft has 24/7 support for all connectors.
6. Monitoring: Built-in monitoring is a click away and with no extra costs. Get alerts and track trends in the usage of your connectors.
7. Robust: Built-in robustness in retry and throttling of connectors exist out-of-box.

Conclusion

In conclusion, integrating Microsoft Dynamics 365 with Azure Connectors gives businesses a powerful platform to automate their business processes, streamline their workflows, and enhance their overall productivity. With the use of Azure Connectors, businesses can easily connect Dynamics 365 to their existing applications and services, allowing them to automate workflows, streamline processes, and enhance the overall efficiency of their organization. The benefits of using Azure Connectors with Dynamics 365 include increased efficiency, improved data quality, enhanced collaboration, and customizable workflows. We can do all this for a fraction of the cost and with full support from Microsoft.

Become an expert using these connectors, and you will be a valuable resource going forward in any size digital transformation project!

Waking up at 6.17 AM was not by accident. My personAId Xris (pronounced Chris) awoke me at the end of my rem cycle, knowing that I would still have enough time for my obligations this regular Tuesday.

I frown briefly thinking of the nanosec pricing for the publicly accessible quantum computing services. Gathering daily insights from my 17 petabytes of demeanor data (all my past data interactions) costs about 3,23 nanosecs. Xris requires these ongoing data analytics to have the insights needed to act as my assistant and business partner.

Tuesday has physical exercise first on the agenda. I put my AR glasses on and start running down the familiar island trail. Conversations from ongoing projects are hovering transparently and steadily in front of my field of vision. I whisper commands (voice commands only audible to Xris) to give me the necessary updates.

I get a call while listening to Xris highlighting the sentiments from today’s conversations and essential excerpts. The priority call from my manager is allowed directly to me without being greeted by my conversational bot. I have been running for 20 minutes, so Xris modulates my voice to disguise the shortness of breath.

I’m asked to join an urgent meeting to discuss some last-minute project changes. I realized 5 minutes into the meeting that I was merely there as a listener and did not have any input to share, so I switched to bot-mode despite company policy. The steady increase of non-physical meeting participation using bots as stand-ins has encouraged many enterprises to enforce meeting ethics disallowing unannounced botmodes.

In bot-mode, Xris can track conversation and sentiments to foresee possible required participation on my part. I would get an early warning and a recap before rejoining the meeting. Instead, I get into the shower listening to the last episode of my favorite podcast.

This might all sound like sci-fi, but all technology envisioned in previous paragraphs, are in fact, very much a reality. But not yet part of mainstream applications.

My ambition was to get you thinking about the art of possible. The only way for this to ever become applications sought after is by implementing them into our next project and it all starts in your mind.

You don’t need to be a large enterprise to utilize the latest conversational intelligence IVR solutions from #Nuance with your #Dynamics365 Customer Service application. Nor does it require a hundred plus customer service agent organizations to reach break-even enabling virtual assistants.

We might not get Xris today, but we do get an experience that will allow organizations to spend their money on development saved from the reduced handling costs covered by the self-services that virtual assistants provide.

For all geeks out there like me, check out Microsoft’s development framework investing in supporting virtual assistants in your environments. Why not make your next customer experience project, whether internal or customer-facing – include a virtual assistant in the teams channel answering all those reoccurring queries?

https://microsoft.github.io/botframework-solutions/index

https://docs.microsoft.com/en-us/microsoftteams/platform/samples/virtual-assistant

Hoping to see many examples and ideas in the upcoming #MicrosoftBuild

Don’t forget to register!

https://register.build.microsoft.com/

#theartofpossible

When integrating Dynamics 365 CE (Customer Engagement) Platform which is based on Microsoft Dataverse with D365 FO (Finance and Operations) we have a few options.

The suggested Microsoft method is to use Dual-write if we require synchronous replication of data between those two applications. Dual-write has a lot of out-of-box capabilities two accomplish a functioning link between D365 CE and FO – but in my experience, this works only well where the use case does not involve a large number of transactions over a short time frame.

The integration templates include a lot of different tables and there are many standard integrations out-of-box for both D365 Sales and Field Service. But as soon as you wish to adjust and add then you will immediately hit various challenges.

You will learn that the Dual write is still a very immature product and has still bugs that you will need to understand and create workarounds for. You will realize that the GUI has limitations and you’ll learn how to deal with those. You will appreciate that many of the D365 FO entities do not support Dualwrite because it lacks the possibility to access data via the odata web API.

But when you finally overcome all these teething issues I believe Dual-Write is struggling with – then you need to appreciate the following do’s and do not’s that Microsoft has manifested.

Out from our previous enterprise definition – we would not see many use cases fitting the above list.

Especially you would most often have multiple instances for D365 CE.

But even if you do fulfill the above list you’d soon hit a wall in terms of large volumes.

Not because dual write has issues dealing with large volumes – but for the fact that I would always try to avoid synchronous solutions for large volumes. I’d always try to design large transactional flows using a decoupled model. There are multiple scenarios where data can fail to (CRUD) create, read, updated, or deleted. We need to take this into account and have ways to deal with these failures and allow the users to make guided decisions to rectify data and proceed.

To analyze and rectify issues with data synch using dual write is truly a technical administrative task.

In addition, we need to consider the API service protection limits with Dynamics 365 CE. Dual-write does not consider other sources or logic for changing the same data and therefore it is a pain to have centralized control of all the CRUDs.

So still as of today 29th April 2022 – I would use other technologies than dual write in any given enterprise scenario.