Відео

Microsoft certs are offered across a variety of regions and languages. Take a look at the exam websites.

Not directly, though given that this provides a local copy of the assets you have backed up, with some extra work you could get them onto another cloud. There are a variety of cloud based tools to facilitate cloud-to-cloud migrations that would probably better fit your need if migration is your objective.

I think the key point is that most business problems aren’t “hard problems” in the sense that o1 or o3 should be pointed at. They are fairly easy problems with very specific context requirements. The addition of some of these new features make getting this contextual information easier.

@@brightideasagency I believe the answer is that it depends. Most businesses have straightforward processes but often lack understanding of how they can impact one another. Why are some people always busy while others are not? As businesses grow, they often become more inefficient. This is because the inefficiencies also increase in size and impact, eventually slowing the flow of finances and halting an organisation. Leadership in these scenarios consider using AI to help 'fix' these issues by taking the existing process and 'speeding it up', but this is often not the approach that yields the best results. So what is the answer? Donella Meadows authored a book titled *Thinking in Systems*, which explores the cause-and-effect relationships within interconnected processes. It emphasizes concepts such as the use of pull-based systems instead of push-based ones, along with other strategies that encourage processes to be more mindful of their surrounding contexts.

@@DavidROliver this is probably true. However, the issue still comes back to context in my opinion. The reason a lot of change fails is through not correctly understanding the context of the problem - consultants chase down what managers think is wrong rather than underlying problems. This same limitation exists with AI. You could give o1 a wonderfully thorough run down of your business' inefficiencies and ask it to design better processes, but if the foundational understanding driving that dialog is wrong, the AI's ability to solve problems - whether hard or easy - is irrelevant. My greatest focus for business AI is where are the opportunities to give it more context? IMO a business version of Microsoft's Recall would be one such opportunity. AI that can watch you work coupled with a massive context window is potentially a lot more powerful than a frontier model supplied with questionable context.

@@brightideasagency agreed ... I wrote this on another post elsewhere. "Start small by implementing a few basic processes and features. It may be a new team, so it will take a few cycles to bed in. Gather your requirements carefully, paying close attention to safety, security, and regulations. It's also essential to consider your budget, time, and resources. The budget consists of two components: the capital required for development and the operational costs needed for maintenance and optimization moving forward. A crucial part of gathering these requirements is defining what success looks like. Next, think about how your new processes integrate into the broader ecosystem of your business or your customer's business. Understand the context, identify dependencies, and clarify the expected outcomes. Clearly outline what actions individuals will take at each stage; a sequence diagram can be helpful for this. Test what you have created, and start by automating these tests to ensure thoroughness. Your measure during testing is whether it meets the success criteria gathered with the requirements. Iterate! Work on the next version without cramming everything into the first release. You want to minimize the time between customer feedback and delivery. Conduct your development in continuous cycles, showcasing new features and improvements each time, paying close attention on how they must work together. Always ensure that your solution is functional and deployed before focusing on optimization. Where improvements or automation are needed is not always obvious, but your customers will provide valuable insights that should inform your improvement roadmap." I'm not really much of a consultant, I still prefer to build things!

I just tested it and it works!! I actually just grabbed a thumbnail from a png in OneDrive and skipped all the other steps but this is great! I could find out for myself but I'm curious if you know, does the create file action create a persistent file or does it just exist during runtime (or 6 hours as you mentioned)?

No. Synology’s software is designed solely for their products and I don’t believe the core of it is open source

Thanks for watching

That is how calendly works. It integrates with your existing calendar exactly as Bookings does. It is an app for booking meetings onto a calendar, not a separate calendaring app.

I'm sure lots of people with the right use cases are subscribing to ChatGPT Pro. I'm not one of them, but my use is not what this is intended for. The question is, how much value does it deliver, and for some niche users, the answer will be far more than their $200/mo investment.

​@@brightideasagency probably just doing a video as you did, but besides that, 200$ for unlimited, when it could have been for the 20$ plan, is crazy, in my opinion. Why would I pay even 20$ for limited access, though? If the 20$ was unlimited, I'd probably pay for it once. But yeah... 200$ is not even going to help them or a good way to get us to pay for that. Sorry if I sound rude or anything

@@brightideasagency I don't mind to sub for 20$ but people won't pay 200$ for a month

@@soul8648 I think you're missing the point a little. If you have no use case for ChatGPT Pro that makes sense for you or gives you value, then I don't think OpenAI cares if you buy it or not. However, if you currently have ChatGPT Plus and you need more because you hit limits, Pro is there as an option (as if Teams, and direct API access if that makes sense). Just because it's most expensive doesn't mean it's best, it's only best if the features that are making it more expensive are valuable to you. Now, if they would be of value to you, but you don't want to pay it, that's ok too. I'd enjoy driving a $100,000 sports car, but that doesn't align with my other financial priorities.

@@brightideasagency sorry about that lol, I'm not a native English speaker, so it's a little hard for me. But I get your point, though; it's worth it if you either make money out of it or use it for something big, right? So I use it every day for anime roleplay aka rp. will it be worth paying 20$ a month? Or just the free plan?

Thank you for sharing this interesting use case. How are you finding the live video so far?

I'm not sure of any that shows o1-mini has changed. Are you seeing issues with it? It is a smaller model than o1, so we would assume poorer results across many domains.

​@@brightideasagencyI've noticed that previously the 01-mino was efficient at tracking my conversations and managing my code requests comprehensively. However, recently it struggle with recalling details. For instance, when I ask for my C++ code in sections, it used to provide all the necessary parts, but now it often delivers only incomplete segments. its like i am using an old model now

I don't think the intended audience of niche users finds the pricing so ridiculous. Some people find a $50+ per month Microsoft 365 E5 tier ridiculous if you can buy Business Basic for ~$40 less, but those businesses that rely on it see the value.

@@brightideasagency if you a business yes, but for just normal users the price is excessive.

@@DavidSmith-uq4mj yes, and nor is this tier aimed at normal users. It's aimed either at users with specific higher level needs in terms of reasoning, or to users who use ChatGPT so much they are limited by the already fairly high ChatGPT Plus usage limits. If neither of those conditions apply to you then no one is suggesting you buy a $200/mo subscription.

Thanks for watching. You learn skills to drive a car. You learn skills to use power tools. You even learn skills to play video games. Why should Copilot be the magical technology that just works without any need for learning? It doesn't make sense. And, on the data front, unless you think Microsoft violates its contractual agreements with customers that include the world's largest corporations and governments, it is using your data exactly as they state. Read the terms of service and privacy policies, and if you're uncomfortable with them, stop using the service.

@@brightideasagency Thx, I wish I could share in with raving reviews. I've gotten different help from other sources that broke instructions "for dummies / beginners" down so I could understand better. I was only sharing my personal experience. I won't be using Microsoft's Copilot any longer.

If you are gaining $20 or $200 or more of value from using ChatGPT, why is it greed for OpenAI to charge that amount? The question on your side must be whether any investment is paying you a return. The only way this should feel greedy is if you're in the category of user who just wants ChatGPT Pro because it sounds better than ChatGPT Plus, but if you have no use case for the extra features or extra compute, you'll get no benefit from it.

I've been using chatgpt daily since 2023! And from my observations, after they made a subscription for 200 dollars, the version for 20 dollars began to work worse, in my opinion, they cut the size of the context window.

I don't see Sora as a compelling reason for the expense, it seems more like a toy to experiment with right now rather than something that can be deeply integrated into value-adding workflows. I'm sure there are lots who will try it, but I question whether most would get $200 of value on the video side versus just a subscription to an existing stock footage service. As the length of video clips increases I'm sure this will change.

Hey, I would like to pay you half of the subscription to share the account, that way is less expensive for both of us. Lmk if you would be interested. Thanks

Thanks for watching!

Thanks for watching!

You are welcome.

Thanks for watching!

Thanks. As we see more of these features become fully available it’s likely I will return to how to use them in more detail.

If your objective is to work live off of the NAS rather than as a backup, I’d suggest looking at the Synology Cloud Sync app versus the M365 Backup. But bear in mind that synced copies of the same file are not technically a backup so does not serve the same purpose.

Thanks for that feedback. Is it having captions generally that distract or just how they look on this video? I used a new process to add these.

Thanks for watching!

Thanks for watching. If you need help with your adoption project, I'd be happy to have a chat: calendly.com/nickdc1/intro-meeting

Thanks for watching. I think the most important point to remember is from M365 Copilot's introduction until a month or so ago pretty much everything under the umbrella of "Copilot Extensibility" was a preview product. Much of that never went GA *because* of the problems, and with agents, in my opinion, has been replaced with something better.

Thanks for watching. While I do sometimes share videos designed as a walk through, the purpose of most of my UA-cam content is to focus on the "why" of the technology or to consider new opportunities. I am a consultant who assists businesses on digital transformation and AI adoption, not a system admin or a daily maker, so there are far better qualified people to look to the content of if what you're looking for is an end-to-end tutorial. If you want to learn the opportunities from new technologies or how they might fit into a more holistic business view, I'm happy to help.

I think the key consideration of the differences here is that with Power Automate you build a track and your process is a train - it can only go where you've laid track and only in the order of the stations you build along the way. With an Autonomous Agent, you build a multi-lane highway with guardrails either side, and let the agent change lanes when it needs to. Power Automate + LLM is still only calling that LLM when you told it to, in the order you set, and with the inputs and outputs you defined. Where you need a track that always gets followed to a destination, then just give the Agent access to do whatever you need in Power Automate - give it a flow that both seeks approval AND sends an email, rather than giving the agent capabilities to seek approval or send an email and relying on it to do both. It depends on what you need to achieve and how critical the sequence of your steps are.

Yes, custom agents are bound by Copilot Studio licensing rather than Microsoft 365 Copilot licensing. However, starting in December we will be able to apply Azure-based pay-as-you-go billing to Copilot Studio, so it'll be possible to get started with these tools without committing $200/mo upfront.

@ what’s a definitive “custom” agent one that uses another model other than copilot? Microsoft is so confusing with what’s a feature and what will need a license. Will any autonomous in the copilot studio within the M365 Copilot plans be incorporated? That’s a high premium for small businesses

@@rezaebrahimi9807 I feel the same and have the same question. It’s not entirely clear in MS licensing guide for Power Platform, but my interpretation is the ‘unlimited’ messages only apply to Copilot Agents built within M365 Copilot. If you are using the full Copilot Studio, so standalone / autonomous agents, and you have a M365 Copilot licence, those unlimited messages won’t count and come out of those $200 month 25K message subscription, unless I’m mistaken.

Two different types of agents: declarative agents extend Microsoft 365 Copilot, and whether you build them in BizChat or in Copilot Studio their use is included in your M365 Copilot license (but they are not available to users without that license). Custom agents can be built in Copilot Studio or other pro-code approaches. If you use Copilot Studio for custom agents, then Copilot Studio licensing applies. If you build a custom agent outside of Copilot Studio then you will just incur service costs for whatever platform you choose to use. However custom agents can be used without a M365 Copilot license.

Bookings works fine with non-standard calendars in group scenarios. Calendly makes this a little more complex as by default the calendar it connects when you login to M365 is you r user calendar. I imagine if you set login credentials to that shared calendar account, it would work just fine, but it's an extra layer of complexity. Additionally, because Calendly by default is user-centered, you would need to remember to demarcate the booking types assigned to that shared calendar. Whether that aligns with Calendly's terms of service, I'm unsure.

Thanks for watching.

Check this out: learn.microsoft.com/en-us/sharepoint/understanding-permission-levels

No, the people booking appointments do not see the appointments of others, they simply see if a timeslot is free or not. The purpose of Bookings is a one-to-many or many-to-many secure ability to give others access to put appointments on your calendar, it's designed to have multiple client users for each calendar.

I’m definitely keeping both. In my view the integration of AI into Microsoft’s existing line of apps is a positive for them, it creates a transition opportunity for workers and processes that never step outside of those apps. And while the computer vision aspect of AI integration to just follow along with what’s on screen is personally appealing to me, I remember the outcry about Recall doing this to store to a local database, let alone sending out to a cloud LLM.

​@@brightideasagency As someone who designs enterprise systems for a living, the Recall function did not cause me any alarm because logging, monitoring, and observability have been fundamental aspects of software and operating systems for quite some time. A SIEM (Security Information and Event Management) system, such as Splunk or Microsoft Sentinel, is used to analyze the data and messages generated by programs to identify any unusual activity. This approach is often how complex security breaches are detected when traditional antivirus solutions fall short. SIEMs have been one of the systems that have led the change in the use of AI having used Deep Learning models way before Large Language Models were invented. The security industry has viewed monitoring desktop activity as a logical next step. While Microsoft wasn't the first to propose this idea, the way they announced it with such fanfare may have been the mistake. Usually security announcements aren't given such a spotlight.

However, Recall is not a security tool, nor does it currently have any broad enterprise capability for the business to utilize that Recall data. This was probably a miss on Microsoft’s part - Recall as an enterprise compliance tool would get people a lot less heated I think.

I think the big reason would be the integration. Even if Teams is Slack minus 2%, is that 2% gain worth managing another platform? I’m not a Slack user, so I cannot comment on exactly what those differences or similarities are beyond the visual ones.

Thanks for watching. No. If your calendar is marked as busy, then Bookings will not double book you. The solution to this is that the sessions you don't intend to attend should be marked as free - they will stay on your calendar for your information, but your availability will be bookable.

I think we see them throwing *something* at the wall to see what will stick. Though I do feel confident the strategy for agents will become more broadly aligned - we can see that in the UI. There are at present definitely better options but not necessarily ones that integrate with M365 with more ease, and it’s this whole package I think is important to many. However, your concerns are very fair.

Thanks for watching and for your feedback on this video

Thanks Andreas.

Thanks for watching!

If by existing agents you mean previously built custom agents/Copilots, then no, certainly not currently.

The challenge with this requirement comes down to wanting the last response as the content of the next response (not really what LLMs are so good at). In a custom agent (in Copilot studio) you can use an action step to do this, though depending on exactly what you’re looking for it’s not going to be exactly a copy of the content. Same for declarative agents, though you run into the same issue. It’s not going to be 100% copy and paste like Copilot pages are. Though soon we’ll be able to call declarative agents inside Word, so that might help with this requirement?