AI In ITSM: Challenges, Opportunities and Security Compliance at SKILup Days 2024

Hello everyone. Uh, welcome to Scale Up Days, uh, uh, you know, session. Uh, hope you're enjoying the, uh, session from the video speakers.

And today I'm going to bring about AI and ITSM, what are the key challenges, opportunities and security compliance. So, let's see the agenda for today. Uh, I'll first of all introduce myself, uh, briefly, uh, and then walk you through what are the key market trends and what is the opening remark for today's session?

What is it? TSM and it, I, in general, we're not going too much in depth regarding to it, i, l and itts M, but just wanted to provide you what is it, tsm and it IL And what are some of the challenges with the legacy systems, which we are having? What are then moving on is, once we onboard ai, since it's the, you know, market and trend of ai, what are some of the use cases of AI in, I TSM with that.

What are some of the benefits and opportunities we see in AI under IT sm? And finally, what are the challenges? You know, we see new skills, we see benefits.

Again, not all technologies are, as, you know, boon, some as been as well. So what are the challenges of ai? And finally, uh, will wrap up with some of the people, process, technology, you know, what is the best mechanism to onboard, uh, a tool, uh, you know, what are some of the, you know, consequences.

And towards the end, I will just provide you with some of the closing remarks and we'll conclude our session. So with that, let me introduce myself, uh, myself. Isha, I'm a cybersecurity.

I'm the professional. I'm also a technology and educational consultant. I have various certifications, uh, be it in cloud or cybersecurity.

Uh, under cloud. I have, uh, you know, been working with Azure on Oracle Cloud. I've been a certified scrum master.

I'm a certified DevSecOps leader from Practical DevSecOps. I have six plus. Uh, I research papers on Kubernetes, uh, uh, software defined networking, uh, cybersecurity, uh, you know, and, you know, various other white papers.

Uh, I've been, uh, you know, contributor and, uh, been speaking and sharing my, uh, you know, my views as a thought leader in on various topics, uh, including cybersecurity, cloud, uh, agile, smart cities, startups, financial operations, s experience. Been, uh, I worked across several industries ranging from telecom com and company to mortgage, to consulting to education sectors. I have both a master's as well as a bachelor's degree, and I'm affiliated to various, uh, organizations.

And, uh, you know, uh, you know, part of various, uh, committees including DevOps Institute. Uh, I'm, uh, part of IEE from past, uh, more than, uh, close to a decade now, albeit, uh, ie. Canada, IE region.

I'm also community builder over here at Canada DevOps Community of Practice, where we host several events, uh, you know, and have a community and various meetup chapters, uh, where in North America. So with that, let's get started. And, uh, first look is what is the market analysis?

This is how we, you know, we see, okay, it's a trend of ai. Uh, chat is been there from past two years. Uh, you know, OpenAI has launched several versions of it, and several companies have predicted the use of ai, uh, from now, uh, from now until, uh, you know, 2030, uh, be eight, you know, Gartner, where we see that 70% of employees are using ai.

And, uh, 45% of executives say that, you know, chat has increased productivity, be eight content development, or be in it with HR or, you know, or marketing technology. And so, uh, some others consulting firm like p WC have also, uh, predicted that that is trillion of dollars of will be used on AI by end of 2030. So there was some free predictions, okay?

AI will be, you know, by 2028, we'll see up to billions of dollars of use on AIOps. And now we have almost, we are in the, you know, closest to 2025. And we are seeing the rise of ai, uh, you know, a lot of AI models, a lot of, uh, companies are implementing and using ai.

So AI is blooming. So let's see what, how it, uh, you know, welcomes itsm. And some of the opening remarks over here is, uh, you know, ai, TSM, sometimes it's called as artificial intelligence service management.

Some people also called as AI ops, uh, artificial intelligent in, in IT operations. Some considered as into itsm. Some consider in tool as sm, where's enterprise service management?

And where we see that, uh, it is a growing solution for majority of IT leaders including, uh, you know, all c c-level executives and as well as developers and employees, uh, where they can see, uh, you know, progress in the business growth, faster delivery to, uh, solutions, as well as, you know, positive feedback review from their respective customers, be it whichever sectors they are in, uh, regardless of finance, hr, mobile application, or be it also a software for, uh, you know, home Mortgage Association. And, and with that, uh, it's also, uh, sales as a, like a multiple technologies and platforms and applications, uh, generating, uh, uh, trillions of amounts of data, uh, including, uh, you know, large amount of volumes including, uh, with, uh, logs analytics metrics. And this has to be seamless where, uh, you know, it's user friendly as a end-to-end flow is, you know, uh, seamlessly integrated with other applications, and we get, uh, end results, uh, you know, as smoothly delivered as well as, uh, their effective solutions as well.

Let's go now, little more in depth in I ts. So I tsm, uh, is, as I mentioned, it's technology service management, and facilitates the objective of organizational IT service to robust maintenance of industry best practices of it. IL now we'll see a new term, ITL for some of the new users as some of the new audience over here.

Let's take a bit pause. I have more to come. What is ITL?

But ITSM is essential in a lot of organizations to keep the IT operations, running, meeting customer expectations, and driving business growth. So what is I-T-I-L-I-T-L stands for Information Technology Infrastructure Library. Uh, in 2019.

It, I launched their version fourth. Uh, I think ITL is there in the industry for more than two decades now. And it addresses modern, IT challenges, especially when our AI has been, you know, tremendously, uh, in using the market.

And the new ways of working has been changed post covid. So it addresses the modern challenges, including agile, DevOps, and lean principles. And it also gives a approach and a framework to service management, collaboration, and a flexibility asset management and co-creation.

Not going too much in depth into ITL, but it's good to see the seven principles under version four of it. L starting with focus on value, start where you are, progress it with feedback, collaborate and promote visibility. We have to think and work holistically, keeping simple and practical.

And towards the end is, and the last principle is optimization and automation. Automation. So it's good to know what is I tl because it's a, you know, subset of ITSM, uh, people usually forget or, you know, mis present.

Uh, I tl but, uh, that is a driving factor for ITSM as well. So, uh, we have seen ITL, we have seen I dsm, but little bit going more in depth into I tsm so that we know, okay, what are the, some of the drawbacks and once what happens when AI will be integrated or when we onboard AI related tool through the SM ITSN. So ITSM is a process and practices that help organization manage and improve IT services, uh, so that it makes business goals, uh, business objectives, and, um, and it's, uh, doesn't defeat the purpose of what is going to achieve, which is for the end users and for the end customers.

Uh, again, uh, taking the, you know, the role of, uh, budgeting and cost, it's very important to dictate what is the overall budget, uh, of companies, uh, you know, uh, overall budget of companies, uh, policy, uh, bringing through procurement to supply chain, and also to meeting the IT department's, uh, you know, uh, quarter results. So a lot of tools are there under, uh, ITSM. And, uh, with that we have seen some of the drawback and, uh, you know, related to ITSM as well.

Like, you know, there are like tons of tools available in the market, uh, tons of, uh, resources out there. But let's go back to legacy system. You know, when we say legacy system, it, it directly refers to system, uh, that does not demonstrate, uh, you know, any new tools or any new technology, which can be, you know, uh, integrated seamlessly.

And, uh, that can make the life for customers and the employees as well. And with, uh, legacy, I pss, uh, there is a continuous need for like, you know, manual supervision, manual approval, manual, uh, permissions, uh, you know, a lot of times system crashes, a lot of time the outages. So that are some of the things we have seen, uh, whenever.

I also work with a lot of customers, uh, and, you know, when we, you know, in my previous engagements of projects as well, like there is this traditional tool they have onboarded, but regardless, uh, it's broken, uh, just so, uh, purpose of regular ticketing system. And some of the key challenges, uh, you know what, like I, I tsm, uh, is higher cost, uh, you know, lot of, uh, you know, overwhelming and missing solutions within that integrated solution. Uh, of course that is, you know, less scalable.

And what it does is just does the repetitive task. So these are some of the, you know, problems with the legacy system. Now, if we consider just, uh, uh, on a legacy IT service task, and one of the challenges we have seen is massive loss of employee productivity, lot of increased frustration within the employees.

Uh, when we want to see a article or knowledge base for a newcomer or a, you know, a new internal, for example, he or she has to go to tons of resources, uh, across the, the internal portal website. And it's a lot of time consuming. And at the same time, if we have to generate a new ticket, lot of forms has to be filled, and it takes, you know, weeks to months to get an approval as well.

Uh, a lot of, you know, troubleshooting issues, uh, there are a lot of manual workflows. There's less automation. So for a change ticket, lot of people, and a lot of managers have to approve it, you know, manually.

Uh, that means manual workflows means like, you know, there is high prone to human errors. And when there is approval of tickets, uh, that means you'll get notification. And you'll see in the emails, there are like tons of notifications coming just because, you know, there's less, and, you know, not good integration as well.

So these are some of the biggest challenges we have seen so far, uh, in, in traditional I dsr. And, uh, what we have seen so far is like a similar workflow, uh, which you can see on your screen. Uh, and most of the companies, and, you know, small, medium businesses have is when an employee wants to request and, uh, you know, access to X number of, uh, any like software or something, uh, he'll call in, you know, agent, maybe that is an IT guy, uh, you know, with, uh, or teams or Slack or just, or a mobile phone.

The service agent will then create a ticket. He'll go to Okta or the, whatever your authentication is as an admin. Uh, he'll confirm Okta Access, he'll create a profile on Okta.

He'll go back, uh, to the manager, uh, if there is a need for approval. If not, then employee will receive a confirmation, your profile will get added. That will be a two factor authentication.

You, uh, and then you'll receive a note, uh, notification, and then your access is granted. So this is a big workflow, which most of the companies has it. What if the service agent is on leave?

Uh, and if you need an access immediately, your of work is then taken to more than two weeks. Uh, if you do not get an approval, you have to wait. You have to follow up, you have to give ticket number.

So there is n number of problems with the simple workflow. So what now, AI comes into picture, uh, and if AI is integrated, uh, you know, now we have several models, um, being to, you know, lama, to meta to Claude, and a lot of LLM models. So let's see if we have a good designated AI and ML team, uh, you know, good insurance team.

And we, you know, we have good skills of people. So we, let's, uh, now see once artificial intelligence and automation comes into picture and how's the workflow look like? And on your screen, we can see this is a typical workflow.

Uh, what can happen when, uh, AI is integrated, I know, uh, through our incident management system, be it like, you know, JIRA, your ServiceNow, or your manage engine. And, uh, this is how, uh, it'll be integrated throughout the IT department. Uh, the IT TSM tools will be, you know, push notifications through emails.

That will be monitoring tools like Splunk and, you know, uh, SolarWinds, which will do like, you know, uh, realtime monitoring and runtime monitoring. Uh, developers will, uh, will have an a p integrated within the IM tools, which will also do the analysis, uh, based on the previous records of the tickets. So that way the, you know, amount of data coming in is less, uh, your history has been stored.

And we have a clear idea about the reporting your dashboarding. We have designated team set up for this, uh, the new ITSM or the enhanced version of I TSM tool. And most of the process is automated through scripts, uh, written and Python, uh, you know, in other languages.

And a lot of the repetitive tasks, which was happening of manual work was happening. Now it'll be automated. So, let's see.

What are some of the use cases of AI for IT sm? Uh, you know, the integration of, uh, chat bots, uh, rapid enhancement of knowledge basis, uh, content management through automated approval and, and scaling up the service request management. And as well as, you know, there will be streamlining of repetitive tasks of some of the common IT issues.

So these are some of the best, uh, six use cases, which we have, which we see in, you know, in AI, ITSM or AI ops. And, uh, with that, it's not just, you know, we are integrating, uh, AI within, uh, just IT department. The use cases can be for several, uh, other, uh, areas as well, be it your hr, be it facilities for your finance, for your onboarding team.

If you wanna go and, you know, wanna check your pay stubs or 4 0 1 or for example, or you wanna different go and policies, or if you wanna relocate or migrate to different countries for care, US or, uh, Australia, you have, you can go directly to HR where, uh, rather than, you know, emailing or just, you know, waiting for their reply, you can, if a feature is integrated, a chat bot has been installed, uh, then, you know, just writing a two or three prompts, you can get, uh, direct responses, uh, with that, we'll also, uh, get, uh, you know, links to the resources. So that's where we see that, uh, you know, AI is just not in it, but it's across all departments. If you, uh, want to see, you know, uh, get access, or if you wanna see what is, uh, you know, my laptop is not working, uh, we have pin issues.

So that's where you can see direct responses when you install, uh, you know, a chat bot or, you know, there is a seamless integration within their request ticket as well. Some of the key benefits and opportunities of using generative AI in ITSM is automated ticketing and, uh, issue resolution, uh, in a proactive analytics of proactive problem, uh, management. Uh, if you want to get, uh, you know, right, uh, how many servers or how many resources are there through tagging of, uh, uh, the best tagging you can get assessed.

You can get, uh, resolution for asset management, uh, smart pricing, uh, of your IT problems, uh, through, you know, your previous history, previous analysis, uh, you have announced, uh, security and compliance, and you have good, uh, area for, you know, knowledge management and documentation. So there are a lot of good key, uh, you know, benefits and opportunities, uh, which we can see, uh, using GeneRead AI and itsm just by integrating the, you know, the best, uh, and use case depending on which business line, which sector you are in. Going now into some of the challenges we have seen, what is ITSM we have seen, okay, what are some of the, you know, uh, legacy systems, some of the use cases benefits, but at the same time, there are challenges as well.

While implementing AI models within ITS and or integrating a new ai, uh, related tool, first and foremost is, you know, uh, experience and expertise, uh, is that fear of jobs as well. Like if, uh, you know, company or team wants to integrate a, a new model into ISM, there is a job displacement of, uh, since it's, uh, you know, you want to integrate and it's, uh, people have less expertise, then that is the first thing that will be a fear of job displacement. Second is, uh, you know, a lot of companies might need to get training programs and change management initiatives needs to be done.

They need to know in and out about the tools in and out, okay, what are some of the, uh, this x, y, Z tools rated how it'll be effective? Um, and third and last most is, uh, and having experience in throughout the end to end system, not just, uh, machine learning, but at the same time, you know, analytics, uh, programming as well. So these are some of the organization changes, which we have seen.

Next going forward is some of the practical challenges in the sense is, is first and foremost is ethical AI concerns. You know, that will be, uh, you know, it has to be transparent. It has to have guiding principles.

It has, it has to go through several, uh, you know, trial and error methods, a lot of dev test environments. There will be times there will be biased opinions and responses. So we have to test the model thoroughly.

There will be chances, uh, or, and, uh, of, you know, data getting breached. So we have to make sure that some of the data we are testing is, you know, is not confidential, and it's just test data. So, you know, to make sure about the privacy and trust, how well, uh, employees are happy onboarding this new tool.

So first foremost, before going the customer, it's important that we get the best user acceptance and how we maintain the customer relationship management as well. And there are times there can be limitations and hazards, uh, related to software, software braking, yeah, model braking, so making sure that, uh, you know, it's compatible, uh, uh, with other, uh, tools and across various other integrations. So there are challenges related to software as well.

So these are some of the practical challenges, uh, um, seen, uh, while implementing AI models with ITSM, some of the technical challenges are, are like, uh, data quality, data volumes, high amount of data. Data has to process at the right time because we are now, uh, seeing the meantime to resolve from, uh, weeks to hours. So that way, depending on data quality, we have to make sure our data security, uh, as well, uh, data is protected.

There is no PI data provided. Uh, when given the response, uh, data is not shared with third parties, uh, it has to, uh, the AI model has to be continuously monitored and improved. A lot of security compliance framework has to be enabled if you are a healthcare company, making sure that AI model is, uh, compliant with HIPAA compliance, IO and this framework has to be implemented, and a lot of governances and regulations has stood also needs to be, uh, you know, made sure that we have the right law enforcement law, right, uh, you know, uh, legal regulations and a proper framework when implementing gene AI models with I ts.

So, let's see. What are some of the next steps for the organizations? You know, with use cases, there are challenges.

So as a, you know, uh, tips and, uh, you have to, you know, educate your employees, uh, by, you know, strategizing risk education program. Uh, if your companies doesn't have a risk and privacy officer, it's a best time to, you know, hire a risk privacy officers. So that way everything is, uh, compliant and you have a proper framework and proper model, uh, by conducting regular AI workshops.

Um, that is the, you know, comes both under engineering plus the HR department, uh, in training and learning and security protocols for ai, uh, creating a risk register, uh, metrics like, or meantime to resolve whether you lay down, uh, what are the risks associated AI models before implementing or going into production, uh, for I dsm, creating awareness via newsletters, campaigns, emails, and, uh, setting up a pilot program to test, uh, through your employees to your trustworthy customers before it's, you know, uh, out in the market. So these are some of the next steps or against need to do, uh, know, before implementing AI under itsm. And lastly is people, process and technology.

Uh, we have seen, uh, what is the process happening? We need to hire the right people. Uh, uh, companies doesn't have expertise doing hiring, training, program, uh, awareness, uh, know lot of, uh, you know, internal knowledge sessions of under AI models and under itsm.

But, but at the same time, uh, it's technology and tools which can break. So that's like, it's not just going and onboarding a hundred of tools at the same time. It's based on your use cases, uh, based on what fits, uh, your needs, uh, in the sense of change management, incident management, knowledge based management, and how AI is capable of giving you output or in returns, be it 10 x to hundred x.

Uh, one of the, you know, wanted to share is one of the live case was, uh, air, uh, which, uh, adopted chat bot, uh, uh, and within two or three months there, it has to shut down the entire, uh, you know, chat bot system because it was giving incorrect responses to the passengers, and which led to a lot of, uh, chaos and a lot of, uh, you know, uh, legal, uh, issues. So that's where, uh, you know, with, uh, boon there is been as well. So making sure that, uh, AI model is trained right, practice, right, uh, you know, right team and right framework is needed, uh, for any of the, you know, tools.

And, uh, with AI TSM tools, there are, you know, hundreds of tools which are enhanced versions, which are there in the market, starting from ServiceNow to, they want a manage engine to Jira service management, but it's, uh, based on a use case and based on the a necessity which, uh, one company has to onboard, uh, within their, uh, you know, IDSM or ESM, uh, management system. And with so many tools, it's like finding a pin from a haystack where we need to go, what we need to do. Uh, as I mentioned, it's based on use case, but at the same time, you have to be mindful about your cost, your budgeting, your licensing and subscription version.

Uh, do it has, it gives out of the box connectors, uh, does it have integrations, uh, does it solve your problem? So these are some of the, uh, know things you need to look when you onboard a new I TSM tools. And this looks like, uh, like a broader picture about the entire ai, ITSM won't go too much in detail, but just wanted to give you overview about, uh, the first layer from the resources to collaboration to the scope, um, you know, bottom line for your knowledge management to, you know, workflow design, this how your entire ai, uh, system would look like.

And with that, uh, we have come to the end of this, uh, session. And my, some of the final, uh, thought words are with some of these, uh, generate AI models coming into picture organization can speed up the process and both internal and external communication and proactively can address issues. Uh, in doing so, the operational efficiency will be improved and also leading to a friendly user IT environment.

And the future we can see over here is to be innovative, but at the same time, we have to be mindful about, uh, some of the pro ethics and practice related to ai, uh, related to security related to compliance, and what are the possibilities and how each one can, you know, unlock new possibilities throughout the ITSM practices. So, with that, uh, thank you for being here. Uh, I hope you enjoyed my session, and if you have, uh, any questions and if you wanna interact, uh, related to ai, ITSM security, feel free to reach out to me on my email or LinkedIn, and I'm happy to have those conversations with you.

Thank you.