AI for Business with BCN

AI Kickstarters; everything you need to know

• BCN

Share episode

Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.

0:00 | 26:03

Harry Busby and Andy James join Pete Filitz to talk through everything you need to know about BCN's AI kickstarters, which are designed to introduce businesses to AI services and provide practical use cases.

The Kickstarters include Copilot Studio Chat, Copilot Power App, and AI Builder; which cover everything from chatbots, automation, and generative AI, providing businesses with a starting point for their AI journey.

Each Kickstarter is a time-bound "taster" program, providing hands-on experience with AI capabilities and practical benefits for businesses.

For more information head to AI Kickstarters - BCN

Thanks for listening!

Connect with us on LinkedIn or visit our website.

Peter:

Hello and welcome to the BCN Podcast. My name is Peter Phillips, and today we'll be talking about AI Kickstarters, what they can do for your business, and how you can get started with AI technology in your business. I have the pleasure of speaking with Harry Busby and Andy James. Hello, gentlemen. Hello. Afternoon. Good afternoon. So, quick round of introductions for our audience is always good. Harry, why don't you introduce yourself?

Harry:

Yep, I'm head of automation and efficiency and doing quite a lot of work with the AI services we have at BCM. Excellent. Andy.

Andy:

I have the pleasure of looking after our productivity solutions team. We use Microsoft platforms like SharePoint and PowerPlatform to create solutions for businesses that try to make processes easier, remove administrative burden.

Peter:

Excellent. Thanks, gentlemen, for your time today. So on today's episode, we're talking about AI Kickstarters. Why does my business need one? And where can I find out more? I think let's dive into the detail. Harry, why don't you give us an overview on the BCN AI Kickstarters and why we've designed them?

Harry:

Yeah, so BCN's AI Kickstarters, they're really designed to start customers on their journey utilizing AI services. So there's obviously been a lot of development in the AI space with generative AI, GPT models, ChatGPT as a kind of commercial product that's available for everybody. And since that's kind of getting traction and people are looking at that, what we really want to do is introduce how that can help businesses and also look at you know different ways that we can help businesses using AI services, where you don't necessarily need to have the full investment of a full project to be able to see what the outcomes would look like. So it's all designed to kind of be short package projects. Um they're engagements that are time-bound to basically meet customer needs without going the full hog and having the full investment of a full project that ends up with a fully productionised product. So it's all about giving a starter to a customer, um, allowing them to see how AI would you know improve their business processes, how can we increase efficiencies, how can we make their business more productive without them having to do that huge investment outlay. So it's really getting things started.

Peter:

That's really useful. And I think um that's what we've seen certainly in the marketplace. You know, businesses have heard the the buzzwords and terms around AI, but it's quite difficult for them to contextualize and actually understand what it means for them and their business. So being able to give them a taste or so to speak is a great way of, as you say, introducing them to the technology and what it can do for them. So let's dive into the different Kickstarters we've put together. We've got a few yeah that we've got listed, obviously focusing, I think, on the more prominent AI services that we're seeing in the marketplace today. Why don't we start with a copilot studio chat Kickstarter?

Harry:

Yep, so this is really around introducing the concept of copilot, which you know people may be familiar with, may not, but effectively virtual assistants, chatbots that work alongside you within the way you work. So there are two kinds of copilot, if you like, that are out there at the moment. The majority is um off-the-shelf copilots. So if you think about Microsoft 365 copilot that's out there today, integrated, built by Microsoft, and put into your applications that you work with today to help you with your day-to-day work. What we're looking at here is building your own copilot using some of your own data. So it's really allowing you, your business, and your business users to interact with your business data, ask questions about that data, find insights about that information, and really building your own chatbot for your own use case and your own business challenge. So, as part of this engagement, we will identify a business process or a use case that fits your challenges. We will have a workshop to gather the requirements to build that. We will then build that, and the output from a customer point of view will be a handover workshop, a handover of a working co-pilot with the data from the business. And really it's it's there to make sure that customers and users can interact with business information without taking one of those off-the-shelf products. So it's building your own chatbot effectively that will work with your business.

Peter:

That's brilliant. And Harry, am I right in saying that's a five-day program? So from start to finish, the customer will get exactly what you've just described there within five days.

Harry:

Exactly. Um, so like we said before, it's uh it's a time-bound engagement, and those five days are designed and structured in a way that you will get something as an output that you can touch and feel, work with, but it won't be a fully productionised product. It's something that could be taken forward by the business to actually make it work for production. But it's all a taster, it's all about giving background to how these things work and also providing something that's that's hands-on for you as a business to touch and feel.

Peter:

So if we just touch on the requirements for this particular Kickstarter, the client needs to have their data in a Microsoft environment, right? Ideally, something like SharePoint.

Harry:

Yeah, so for this one, there are limitations on what the dataset can be, um, just based on the copilot studio integrations for a program like this. So it needs to be either in a SharePoint site or on a using a company website, so we can interact with that and have that as a front end to company website information. Or we can, as part of this, just upload files into the copilot studio directly and have integrations with those documents to ask questions about that. So those are the three data sources, if you like, for a Copilot Studio chat's Kickstarter.

Peter:

I I guess again, to give our audience a little more context, do we have a typical business use case that we could reference? So a client has a particular need, and what would that address? Could we reference any examples where we've done something similar?

Andy:

Yeah, so we've had loads of conversations with a number of people, lots and lots of opportunities. It's been quite exciting. And that's ranged from companies who are in the retail sector and they're looking for a chatbot that can sit on top of their catalogue of products where uh a user internally could ask questions of the chatbot to find out more information and to have a maybe a more curated list of relevant options put together for them. So if you were looking at uh cars that you might ask for a list of the vehicles that you sell that meet certain requirements. Uh and then you can ask more questions about that type of car, and all that information can be automatically generated by the AI. Similarly, I mean that could be used internally for people, but also you could then put that chatbot onto your website so that members of the public and potential clients could then interact and access that information as well.

Peter:

So, really an intelligent search engine that you can interact and engage with as you would a human being.

Andy:

Very much so. And it's that natural conversation where you can ask a question, have a conversation to get the information and not have to worry about, well, what's the keyword for that and where do I search for that and how do I put this search together? As Harry said, part of this program will include a prompt workshop where we talk about how you engage with these chatbots to get the most out of them. But it does give you a more flexible, natural human way of interacting and searching through large quantities of information.

Peter:

Brilliant. Moving on, the next one we've got is the copilot Power App Kickstarter. And this is an eight-day program. Harry, do you want to talk a little more about this particular one?

Harry:

Yep, so following on from you know the concept of a copilot as a virtual assistant, the co-pilot power-up kickstarter is really changing the context of that engagement with a copilot into a power app that's been developed. So, PowerApp developed, your data sits in a Dataverse backend effectively, and you can then start to interact with the business data that's within that Dataverse, those database tables, and have the interface through a Power App that exists. Now, if a power app doesn't exist today, then we can provide templates and we can work with a customer to basically build out a different use case using a Power App and interacting with the data sitting underneath it. Now, this one really works best just talking about use cases and talking about kind of how that looks. But if you think you've got a custom application which is a power app, you have a set of data that's associated with that that sits behind it, being able to interact with it through that power app interface, using natural language as a copilot in that power app, then you really start to kind of utilize and surface up the information that's within that power app ecosystem using natural language. So I think Andy, you might have some use cases that really kind of make this real and bring this to light.

Andy:

Yeah, very much so. We've got an opportunity for things like rostering and staff management. So if you're managing fleets of drivers or shift workers in warehouses, manufacturing, or even in public services, and you have everybody's shifts listed and detailed in Dataverse, then in the app where you manage all those rosters and you you authorize annual leave or you update sickness or you change shifts or you notify people that overtime shifts are available. So all of this business process can happen in one place. We can embed in Copilot to help assist with those things, but also be able to ask questions like what is Andy James working on November the 8th this year and then get an answer, or what are the staffing levels looking like in Woking on this date and then get an answer? So the application itself can handle the business process, but then the embedded copilot interface inside that application then allows you to interact with that information and get you know whatever information back that you need. And the nice thing with that is that everything's in one place instead of I've got an application to do this bit, but then I need to go and interrogate data somewhere else. You can kind of do it all in a one-stop shop. So really powerful.

Peter:

So if I was to articulate this, I guess in layman's terms, the app itself creates a sort of gateway to the data, and co-pilot is essentially your assistant that you're speaking with through this gateway for essentially retrieving and finding information as as you ask it.

Andy:

Yeah, absolutely. Power apps are fantastic for business processes. So if we look at the example we spoke about earlier with all your products, you can have a power app that manages inventory and stock levels and placing orders and everything else. But then you can have a page within that app that is your copilot search for you to ask questions and interact with that information. So the app's amazing for business process and a human interface with that process and maybe some automations, and copilot kind of comes in parallel to support that and gives you a really nice interface with the data.

Peter:

Brilliant. That makes sense. So it's more than just a gateway, the app itself has also a function in terms of helping automate and build some intelligence into that business process, which helps drive more efficiency as well as supports, you know, the information gathering of the user's request. And all that for eight days. So I mean that's phenomenal value for money, so to speak. If you can revolutionize essentially a business process and surface relevant company information quickly and readily to decision makers and stakeholders, that is a game changer when we're talking about what conventional project turnaround times and investment requirements were. Now, we've obviously got a few more to discuss on today's chat. Let's move on to the AI builder Kickstarter. Harry, do you want to give us an introduction to this one?

Harry:

Absolutely, yeah. So um this kind of moves away a little bit from the chat interface with AI, which is what a lot of people kind of perceive AI as an overall. You know, copilots, chat GPT, and having chatbots and information, you know, conversations with bots. This is moving towards embedding AI within automated workflows. So not necessarily you know asking questions and analyzing data based on something that a user is looking for. This is more there's an automated business process that exists. Um, we can give some great examples of that where we've got email attachments and email documentation that that we can automate and analyze, but we're embedding the AI itself into an automated workflow. It's not necessarily a user interface of chat, but it happens in the background, and we can surface up, you know, things like exceptions and information like that as part of that workflow, but it's an automation context, effectively. We're moving from chat interface to a context of an automation. So again, eight-day program where we work with a customer to identify a business process. We look at building out that business process in an automated fashion or as automating as much of it as we can, embedding AI into that. So whether it's analysis of documents, analysis of data within that, there may be some natural language processing that goes on as part of that process. But ultimately, it's a background process that's being automated with AI being a part of the solution. So, again, always the case, best surfaced up as an example and looking at business examples, but it's really real-life value that we can get from AI within that eight-day window to get customers started on the journey and make them think about AI in a slightly different way. Because the services included in this, they've been around for quite a while, but they're only really being surfaced in in this way because of the highlight that ChatGPT has put on AI and other things. So we've got lots of uh different examples. I think Andy, if you can give some real business use cases where we've used AI in these in this way, it will really kind of resonate with what this Kickstarter can bring.

Andy:

Absolutely. I mean, like you say, the other Kickstarters we've got give us an interface to work with the AI, which is fantastic. But if we can remove the need for a person to interact and an automation can just get on with things in the background, then even better. You know, it's all about removing administrative burden and allowing people to focus on the aspects of their role that people need to focus on. So if an automation can do a process beginning to end, then even better. And some of the things that we've looked at in regards to this particular Kickstarter include things like processing invoices. So if an invoice lands in an email inbox, that an automation can recognize that, identify the attachment to that email, confirm that it's an invoice, the AI can read that document, extract all of that information, and then a number of things can happen. When the AI extracts that information, it gives a confidence score to say how confident, if only enough, it is that it's read the information correctly. If that confidence score is high enough, then we can just allow the automation to process all of that invoice request and push it potentially into your finance system. If the confidence level is too low, then maybe we want a person to look at it. And the reasons being it might be maybe it was a scan of a document that's a bit smudged, maybe there's a coffee stain on it, maybe there's a number of reasons that the AI confidence might be low. But as long as it's high enough, we can automate the process beginning to end. And like you said, Harry, we can give an interface to a user to just pick up those exceptions. So if the AI is not absolutely certain that it's read this information correctly, there can be a power app that you can open up, you can see the document that the AI has tried to read, you can see what it's translated that as, and you can just say, Yeah, that's fine, and click a button. Or if something needs to change if it's read an eight as a zero or something, then you can make that change in the app and push it through. Again, although you're there to pick up the exceptions, you're just making that slight tweak, and then the automation can carry on on its way. We've got some others which use a different type of AI, so instead of reading, they're generating, and that can be around generating answers for tenders. So if you're going for a tender and you're asked in this tender what your methodology is around project management, when you put that question into the application, AI takes that question, it categorizes it, identifies it as being a project management type question, and then can look at a catalogue of all your historic tenders, identify all of the project management questions from those, and use those answers from your previous tenders to then generate you a new answer. And again, once that's generated for you in the application, you can make any tweaks and changes to it that you want. You can have someone approve it if needed. If there's a particular key stakeholder that looks after project management who needs to sign off on those type of questions, but you can then start to use your historic information to automatically generate some brand new content, and again just take that administrative burden out, try and speed up processes and release people to do what people need to do. So there's a whole raft of opportunities. We could do similar things with building CVs and all sorts of bits and bobs. Really exciting opportunities there.

Peter:

Indeed. I think it's fair to say every business could do with it, right? There is always going to be a process, a workflow within the business that one could automate and use services like this to streamline and free up people's time to, as you say, focus on the bigger picture and the more important elements for which human intervention is needed. Last but not least, we've got the generative AI Kickstarter. Harry, do you want to talk to us a bit more about this one? Absolutely, yeah.

Harry:

So this is a looking more at some of the more, I guess, emerging tech that's out there. So really looking at large language models, generative AI in general, and providing an introduction to business users and to businesses around what's possible using Azure OpenAI, and looking at the GPT set of models, which are you know the power behind Chat GPT. So if we think about some of the things we've spoken about previously, AI builder, those kind of things, they are packaged models that have been curated and worked with Microsoft. Same from this point of view, but they'll be using the OpenAI models, which really are kind of the bleeding edge, if you like, of large language models and chatbots that can be used. So effectively, what we will do as part of the engagement, again, this will probably be an eight day engagement where we'll have a workshop with the business, we'll introduce generative AI, we'll introduce Azure OpenAI, which is running the OpenAI models on Azure with some added security benefits. So that's always great because uh, given the focus that we have around. Security in AI at the moment. We'll work on prompt engineering and what we'll do is set up customers with their very own chat GPT effectively, which is a Zero Open AI with GPT model attached to it, a playground for the customer to actually engage with. And we'll also look to integrate that with a custom data set for what's called retrieval augmented generation. So this is really similar to the copilot Kickstarters that we had, but it's pushing it one step further into the kind of GPT world, into the chat GPT world, and using the power of a large language model which has a base training data set that understands a lot of information, providing a customer data set or documents or something similar to be able to have a full generative AI conversation with that data set. So the difference really is the model and the AI actually underneath powering the chatbot or the copilot itself, but it's really pushing it to the next level to show what's possible with these large language models and what's available today.

Peter:

So lots of food for thought there. I think it would be good for us to elaborate why businesses should be interested in the Kickstarters and what the benefits are. Harry, did you want to cover some of those off?

Harry:

Absolutely, yep. So as we mentioned previously, it's looking at a rapid introduction to AI, generative AI, and really kind of getting users or business users within organizations to have an understanding of what the possibilities are, be able to get their hands on some of these services and some of these capabilities that are out there today and are being looked at, and really make sure that people are have the insights into what this can do for a business and what this can do for an organization. So the idea behind these Kickstarters is really you will get something out the back of it that is yours to keep. You will be able to play around with it, productionise it, elaborate on it, make it whatever you want to make it. But ultimately, we will give you all the insight and all the basis of knowledge for you to understand how AI can affect what's within your business. So that's the first piece. Practical application is the second one. This is really where we wanted to focus with these Kickstarters because we've been doing things like this with customers. We have you know solutions that use these kind of services and these um capabilities that are live today, they do a myriad of different things. What we wanted to do was really kind of introduce those possibilities with practical use cases. So Andy mentioned the the invoices, all companies get invoices, right? How you process those today will vary massively, but the overall process end-to-end is receive an invoice, goes into a finance system. So there's something there that we can really point at that we've implemented with other customers that says, here's how you can automate, streamline, and make that process more efficient across any business, basically. So really showing that practical use of these AI services without, and this is another key piece, all of that huge investment that would be required to build out a custom solution from scratch. This is very much a starting point on a journey, but by providing that you know practical use case and the practical benefits of using something that we work with you, you can then take that forward. We can do a full assessment after the Kickstarters have finished on cost of ownership, what would need to happen to take this to a production solution, and really kind of showcase without a significant investment how AI can affect the business. So it's really using that Kickstarter to kickstart um your engagement and kickstart your projects towards using AI in your everyday business. All of the Kickstarters will come with knowledge transfer. So at the end of that engagement and obviously at the start as well, we will have workshops, we will have sessions with you and your business team and the people within your business to make sure you understand what's happening or what's happened as part of that Kickstarter, what possibilities are next, and like I say, what are the steps you need to productionise? So it's really getting things in place to make sure that businesses are aware is the first piece, but also understand the benefits and can touch and feel those benefits without having that huge outlay of uh initial significant investment.

Peter:

Excellent. Thank you so much for that summary there, Harry. I think that's super useful. So I guess just in closing, AI doesn't have to be daunting, right? With BCN's AI Kickstarters, we provide a low risk entry point to explore and pilot AI technology. And there is no reason why you can't get started today with this technology and see how it can revolutionize and change your business. You know, talking about processes, we've just given you some examples there. Happy to discuss with you in more detail what we've done for other businesses in your industry and how to choose the right process for AI. I'd also just like to take this opportunity to mention we've got an upcoming LinkedIn Live event Thursday, the 18th of July at 11. To find out more information and to see the technology discussed in action, please register. We'll post the link underneath the podcast recording. Gentlemen, I just want to thank you both for being with us today and sharing your thoughts and uh giving us a great overview on BCN's Kickstarters. Thank you so much for coming.

Andy:

Pleasure.

Peter:

Thank you. So if you've liked this podcast, don't forget to like and subscribe. Please don't forget to visit our website www.bcn.co.uk for a wealth of information on AI services. Thanks for tuning in, and we look forward to catching you on the next one.