Відео

No (I mention it very briefly). I would still use Function Calling in production code, as the user of the system might just say "Hi", we have wasted a call to the vector store and extra data on the prompt that was not needed to exchange pleasantries with the user. I kept it out of this video to focus purely on the vector store code

Thanks,. But this video actually have some flaws in that things i cosmos end up with duplicates + Semantic Kernel have a new and better way now to store Vectors, so stay tuned for an updated video on this on the RAG topic

@@rwj_dk Thanks for the heads up. Not sure I'll be using cosmosdb. Looking forward to watch the update👍

The new and improved RAG video is out: ua-cam.com/video/OkH9l_4IX4o/v-deo.html

Thank you for the suggestion. No promises though as Neo4j is not a system I'm familiar with... Plan is advanced version against CosmosDB and then something against Azure SQL Server for now

I've now checked and Neo4J is not one of the supported Vector databases in SK: learn.microsoft.com/en-us/semantic-kernel/concepts/vector-store-connectors/out-of-the-box-connectors/?pivots=programming-language-csharp ... There is documentation on building it yourself but sounds difficult: learn.microsoft.com/en-us/semantic-kernel/concepts/vector-store-connectors/how-to/build-your-own-connector?pivots=programming-language-csharp

The API works on Free and Paid... There are a few methods and properties that do not work/are ignored while running on the Free tier (Like List colors and Assigment of members/due date on Checklist items)

@@rwj_dk thanks a lot

There is as such no value using agents over direct calls in simple scenarios, but in more complex scenarios where multiple agents are used, it is a handy way to define agent-specific system messages and control who/what use function calling... So I've just using agent for everything to have those features available if/when needed

@@rwj_dk Thanks. It would be great if you make an video of real time use case of Agents. I believe that would be helpful for everyone.

Conspiracy theory: Main reason is properly to hide it a bit so people do not make as many workspaces (so they hit to 10 board limit and go to a paid plan)

Tokens are a bit hard to count. Here are OpenAI explanation of them: help.openai.com/en/articles/4936856-what-are-tokens-and-how-to-count-them

And here is their calculator: platform.openai.com/tokenizer

Beyond the tokens for the actual words there are also meta-data around them (if it was the user or AI that spoke them)... It is rather complex to understand the exact number so I just take the Token count as accurate and use it to judge cost, rate limits without spending time on their origin

I mention it briefly in the video (that you can have one AI review another AI's output)... A more concrete example would be an AI that write marketing slogans and promotion texts, and then have other AI's rate the text for legal, plagiarism, trademarking etc. and reject the first AI's work (asking it to reiterate its suggestion untill all others Approve the text)

Thank you for the Feedback... I do have a few bonus topics I could add 😉

Bonus 1 Video is out: ua-cam.com/video/7jnDGfKz0oM/v-deo.html (Let me know what other topics you might like to hear about)

Bonus 2 Video is out (All about Token Counts): ua-cam.com/video/jCl2IyYzkB4/v-deo.html

There is actually no need to make the collection and partition key... The SK Package does that automatically with just access to the connection-string 👍... But you are right: I should have mentions the extra setting needed in cosmos before it accept vectors

@@rwj_dk Hmm. It might be how I have created Cosmos then. It fails with the error "Owner resource does not exist" if I don't manually create the database and collection beforehand.

@ThomasArdal 🤔... They could have changed it since I did it a month back.. I did at no point make the collect nor select the Partition key... In any case sound like you got it to work 👍

Glad it was helpful!

The original source is not available - This is the Best Place to Start (The Wiki): github.com/rwjdk/TrelloDotNet/wiki/TrelloClient#webhook-features

That question kinda have two answers... A short and a long one... Short answer is 'No', TrelloDotNet do not have a method to do this... The longer answer is that the raw Trello API have a search Endpoint (developer.atlassian.com/cloud/trello/rest/api-group-search/#api-search-get) but it it have a major issue in that if you add a card, several minutes can go by before this search feature know the card (guess it need to do some internal indexing). A few minutes might not sound al lot, but it make the feature totally unreliable to the point where I have not implemented it (yet) as it brings little/no value. Instead it is much more reliable to get all cards on the board (Using the GetCardOptions overload to only get the name of the card and then in memory search for the title of the card)...

Hmm, I have not seen that happen in any of my moves. Best guess is that there is some error or the target workspace already have 10 boards on a free plan so it can't accept an 11th

If all 9 users need multiple boards, then all 9 need to pay if the boards stay in a single workspace... If some of the boards are not needed for the advanced premium plan stuff they could be moved to another workspace an that way perhaps save a few paid users (being single board guests), but it is all in the details if that make sense or not...

@@rwj_dk Thank you soooo much for the answer! We just have 2 boards in the same workspace, but i think we can probably delete one and create it in another workspace! But we need the premium stuff at least in one of those 2 boards. I'm gonna try implementing that and see if i can pay just one premium workspace with 1 single board and have 8 guests, then another free workspace with eveybody as members there! I'll let u know how it goes xD Thanks

@tuliobenavides301 No need to delete/recreate the secondary board. You can move boards between workspaces: ua-cam.com/video/XKYLfUJN86U/v-deo.htmlsi=FDBh1YNsTEp_Uhnh

I am indeed using the Webhook system listening to board changes... A webhook needs a hardcoded boardId, but it can be changed if you migrate to another board, or you can have multiple webhooks (one for each board) if you wish DoD on multiple boards. The definition of a "Done" card does not as such exist in Trello, so each team need to Define what state is "Done"... For many that is a Done list or a Done label.. for others it is that the card is archived or moved to another board... Once you have define your state-definition, you simply get all cards and in code filter those away. How the whole thing is done in code can be seen here: github.com/rwjdk/TrelloDotNet/wiki/Automation-Engine

@@rwj_dk thank you for your response, (I am the one sent you an email asking for you guidance on a trello project )

@TOn-fx2gr Just gave initial reply 👍

@@rwj_dk thank you very much I will gather more info from my client at 16:00 and I will send you an email after that .

Tokens are personal keys that span across Workspaces, so I'm not really sure what you mean by the mordern way (The token Consent Form Popup that appear when you generate the Token show all current and future boards across workspaces you have access to) [Screenshot: share.icloud.com/photos/020E3-hJNbY6l0poO_hSRBPWw]... The API key is generated within a certain workspace but that is just to indicate the owner of the key (since one key can have multiple tokens from various users.... So I guess the confusion is that a PowerUp is activated on a per board level, but if that PowerUp ask for authentication, it is actually gaining access to all your boards, not just the one the PowerUp is installed on

@@rwj_dk yeah, thanks, I've managed it. I said "modern way" because the last time I created tokens there was no Power-Ups at all, you just typed the link and was redirected directly to the page of creation Token. And If you did it previously, you will still the see your personal token in a separate block (which called literally "PERSONAL TOKEN") below that message about "go to Trello Power ups admin page". But if you didn't - you won't see it even if you will create token later. That's why I was confusing, because they firstly get rid of "personal" word, moved the page of it's creation to another place, and all of that inside the PowerUp item which literally says you have to choose the Workpace it's related too. This is confusing. But hopefully, the created token despite of how it calls now and there it can be found still gaves the access to all of the Workspaces.

Yes, can't see why that would not work 👍

An API Key/Token is not tied to a specific board or workspace... It is instead tied to the user and what that user can see/have permissions to do... So if you generate a token for a user that example have access to 3 workspaces with 18 boards inside those workspaces, then the API key/Token can access all of those... But if you as an example is not an Admin on a board, your token can't do admin things via the API on that board

Thanks!@@rwj_dk

Thank you for the kind words. You might also want to check out this PowerUp I made that make it even easier to get the ids: ua-cam.com/video/es84INLIiKI/v-deo.htmlsi=E94y26C0O_e4PwLp

Yes. You can use GetCardsForMemberAsync (github.com/rwjdk/TrelloDotNet/wiki/GetCardsForMemberAsync) and GetActionsOnCardAsync (github.com/rwjdk/TrelloDotNet/wiki/GetActionsOnCardAsync) respectively... Or for the action you can use the overload versions of GetCardsXYZ overloads that accept a GetCardOptions object where you can specify that you wish to get the actions of the cards included in the Card-object

Hi. PAT tokens have many different use cases (Github Actions, Azure DevOps Build agents, own code, and so on)… The video is for people that are being told to provide a PAT, but are unsure how to get it. Did you have any specific usage scenario in mind writing this?