About a million years ago, we opened a business banking account with Bank ABC. 

At the time, they had two primary types of accounts, personal and business.

So far so good.

Personal accounts had data fields for all the expected pieces of information one might need for a personal account.

The business accounts had all you might need for a business account.  Or so they thought.  Turns out businesses are owned by people.  People have separate information from the business.  No big deal, we can have attachments and notes.  In the note we put the names and details of the business owners and an attachment of their scanned signature cards.

Then Bank ABC merges with Bank DEF.

Then Bank ABCEDF is acquired by Bank XYZ.

Let’s smush the old data model into the new data model.  Great, all done.  Nothing to see here.

A few years later during a random account review it is discovered that a particular business account of a million years has no people associated with it, no authorized users.  There was an attachment of a scanned image with two illegible signatures on it, but no names with the signatures.

Turns out somewhere among the mergers and acquisitions, notes did not come with the business account information from older accounts that were opened when records were mostly analog and a little bit of digital.

A good bank manager found this, found us, and convinced me she wasn’t a scammer and got us to the bank in person to get our personal information appropriately related to the business information.  And then we signed paper copies of signature cards (sigh).

There are a few things to learn here.

Don’t store structured data in an unstructured place.  Build the appropriate relational data structure that you need.

Don’t cut corners when moving to a new system.  It really doesn’t matter that there were mergers and acquisitions along the way, it was a new system.  A simple review of notes on common data tables could have shown there was more data that needed to be dealt with.

At some point this would have become a blocker for access to our own money in our own business account.  Hire good people who can do good work and prevent this from happening and correct it when it does.

If you were in our hands-on lab sessions, we promised links.  Here you go:

Get Started With Copilot in Cloud Flows (Microsoft Documentation)

Copilots.

Large language models.

GPT this.

Open AI that.

The world is full of excitement for AI. 

And the world is full of doom-and-gloom for AI.

While there are many over-the-top scenarios we’ve seen, the most effective uses of AI for everyday business is in small pieces.  Add some AI here.  Streamline some work over there.  Surgical. Specific.  Purposeful.

The best uses of AI happen when combined with critical thinking directly from humans.

At 365.Training we’re using AI both directly for the benefit of the users while they are training, but also behind the scenes.  I guess that ultimately benefits the user, but less directly.

365Ai is your training copilot- While in a course, you can ask contextual questions of the copilot.  The models are of course powered by the large language models, but also grounded with our own information.  We make sure that the copilot in a specific course can answer topic-specific questions and not get sidetracked on other topics. The copilot is relevant and there when you need it.

mydigest- We’ve got a few different places where we’ve inserted AI. 


TL;DR- We’ve manually curated sources (more than 150 of them so far), and we have AI generated summaries of the items in your feed. 

Categories- We use AI to help identify content topics and the job roles that are relevant for the items.

Your filters- you select topics, roles and importance and we calculate a score and combine that with our scoring of items and apply that to make your custom feed.

Your digest email- we apply similar logic as above and curate a custom email on the schedule you define.

We’ve also used some of the Copilot capabilities of GitHub to help accelerate some of the more tedious parts of development to get us to market faster.

How do you AI in your everyday world?

So much great feedback!  Keep it coming.  Use the Ideas page on the 365.Training site or reach out to me directly.

So, in case you haven’t heard, my digest at 365.Training is a place to stay up to date on all things Power Platform and Dynamics 365.  We’ve got over 150 sources for you, all in one place.  My original blog post on it is here.

So, some updates for you.

Email digest from my feed. You can set the frequency of your digest email from none to daily, weekly, monthly or after 10 days of inactivity on the site.  Your default is set to none, so if you want to get an email digest, navigate to the settings page via the gears icon on the ribbon.

You can now listen to our AI-generated TL:DR summaries.


Suggest a new source.  On that top ribbon, click the plus+ and let us know a great source we may have missed. You can also keep track of your suggestions, and the outcome on the My suggestions tab.

Time to consume.  Once you’re drilled into the summary of an item from your feed, we’ve got an estimate of how long it will take you to consume the item from the source.

You can listen to a podcast directly from our site.  Not every podcast provider supports this at this time, but most of the ones we share will have that option for you.

Once you’ve set your filter preferences, and are viewing your feed, you can also now choose what type of content you’d like to see right now.  Use it just like you use our just-in-time my view filter.

Load more items to your feed. When you first get to your digest, we give you 50 items on your feed.  If you make it to the end and want to load the next 50, use the mark all read at the bottom of the feed. That will effectively dismiss all items on your feed and reload with the next 50 items.

Still haven't tried my digest?  Go give it a try now https://mydigest.365.training