AI in project management: from automation to augmented leadership
Artificial Intelligence is changing how projects are planned and run. Watch this webinar to understand practical impacts and leadership implications. Scroll down to watch the session and reflect on your own approach.
Introduction
Project teams are increasingly using tools that support analysis, forecasting, and reporting. These tools influence decisions but do not replace professional judgement. This webinar explains where they add value.
In ‘AI in project management: from automation to augmented leadership’, Markus Kopko explains how AI works alongside recognised frameworks. He connects new capabilities with AgilePM®, PRINCE2 Agile®, and PMP®. The emphasis is on disciplined delivery.
You will learn where AI fits into risk management, scheduling, and stakeholder communication. The session also explores how leadership expectations are shifting. Each topic is grounded in real project work.
Video
Access the presentation here, listen to the audio, or read the full transcript below.
About the speaker
Markus Kopko works with organisations and professionals introducing Artificial Intelligence in a structured way. He also supports the development of strong project management capability.
He is a member of the Core Development Team for the PMI Standard for AI in Project Management. His work links global standards with day-to-day delivery practice.
His training style is interactive and well organised. Sessions focus on practical relevance, from introductory learning to organisation-wide programmes.
Transcript
Here’s the full transcript of the webinar.
00:00:00 Sevcan Valiyeva : Hello everyone, welcome. I’m just going to give a few minutes while people enter. Feel free to write your name and where you come from in the chat. I can see a lot of people from the UK.
00:00:40 Sevcan Valiyeva : We’ve got one person from the Netherlands.
00:00:45 Markus Kopko : Yeah.
00:00:48 Sevcan Valiyeva : Wow.
00:00:49 Markus Kopko : Ethiopia wrong.
00:00:50 Sevcan Valiyeva : Wow, definitely.
00:00:58 Markus Kopko : Very nice. So, we are in a global event right now.
00:01:08 Markus Kopko : Isle of Man, very nice.
00:01:17 Sevcan Valiyeva : Hmm, from Belgium. So, while you continue, I want to give a very quick background. I’m Sevcan, I’m the market executive for Knowledge Train, we are in partnership with agileKRC. We have Markus with us today; he’s coming all the way from Germany. So, thank you so much, Markus. Just before I hand over to Markus, this webinar is recorded; you will receive the recording link next week. At the end, we will have a Q&A session. So, feel free to write all of your questions in the chat, and then Markus will answer them right at the end. Markus, over to you.
00:02:01 Markus Kopko : Okay, thank you. Thank you very much, Sevcan. Yeah, thank you all for joining. Welcome to our session today. We have a lot to cover, so we will start right away, and as Sevcan already mentioned, if you have any questions throughout the presentation and the short demo I will do later on, please reserve it for the Q&A part at the end of the presentation. Okay, so let’s start. We are here to talk about AI in project management, but this isn’t a session about tools or vendor pitches, right? This is about the professional upgrade that’s available to you right now and the shift from administrative overhead to strategic orchestration. So, from spending 60% of your week on tedious tasks like status reporting, meeting minutes, and spending that time on the work that actually creates value. So, we want to make that shift from the tedious work to more focus on tasks that creates value, right? So that means also AI doesn’t replace you, doesn’t replace project managers. That’s a fear I recognise a lot from many people I’m talking with, but that’s not, will not happen, at least not in the near future. But AI, it replaces the administrative work that’s been holding you back. And when you free yourself from that work, you can finally become the strategic leader your organisation needs.
00:04:00 Markus Kopko : So, let’s explore how to make that shift.
00:04:07 Markus Kopko : Okay. Just a second, here we go.
00:04:14 Markus Kopko : So, this is a new value equation for project managers. On the left side, you see commoditised tasks, right? Status reporting, schedule updates, meeting minutes, data compilation, risk register maintenance, all that tasks you know may be better than me. These are necessary tasks, of course, but they are not where you create the most value, right? So, on the right side, you see premium value, strategic judgment under uncertainty, stakeholder orchestration, complex negotiation, value alignment, change leadership. These are the competences that make you irreplaceable. The opportunity we have today is this. AI can now handle most of the left column, which means you can finally spend your time on the right column, and the market reroutes strategic leadership, not administrative efficiency, and AI gives you the freedom to become that strategic leader. Okay, let’s make this real for a moment. I’d like you to answer a quick question. What percentage of your week is spent on administrative project management work? As I said, by that I mean reporting, documentation, scheduling updates, all those things, all the commoditised tasks from the previous slide, and please be honest with yourself. This is for your own self-assessment, not for anyone else. You may put the number in the chat, but yeah, that’s not mandatory. Okay, so let me have a quick look. Okay, a lot of results are coming in, very nice. Thank you for that. So well, no matter where you are today, there’s a path forward. If you are spending 20 to 40% on admin work, you’re already got bandwidth to optimise, right? If you are at 40 to 60%, and I’ve seen that a lot of time here, you are exactly where most project managers are, and there’s a real opportunity to reclaim 5, 10, maybe even 20 hours a week.
00:07:11 Markus Kopko : And if you are even about 60%, well, you’re not alone, and the good news is that’s the biggest opportunity zone. Because every hour you automate is an hour, you get back for strategic work. I have a little bit of technical issue, here we go. Okay, so here’s your evolution path, three phases, right? Phase one is liberation; this is where we eliminate low-value tasks, like status reports, meeting minutes, risk register updates. AI handles the compilation, and you can focus on interpretation. Phase 2 is amplification; this is where AI enhances your judgment and insights. Think risk pattern recognition, plan review, scenario planning, AI surfaces insights you wouldn’t see on your own. And phase three is leadership; this is where you master the human critical zone. The things AI cannot do, like strategic vision, stakeholder orchestration, ethical oversight, and change leadership. So, today’s roadmap takes you through all these three phases, where you are, where you are going, and how to get there. Let’s proceed with phase one. Okay, act one, the liberation phase. This is about reclaiming your time. Let’s look at two big wins here. First big win, status reporting. The traditional way looks like this; you pull data from Jira, MS Project, Asana, whatever your project management tools are. You manually compile it in Excel, you format it in PowerPoint, you add commentary, and you send it to stakeholders. Time investment, 2 to 4 hours per week, or even more. The AI-augmented way looks different, AI generates a draft report from your project tools, and you review it, you add strategic context, and you can focus on interpretation and not compilation. Time investment, maybe 20 to 30 minutes per week, that’s a 70-to-80%-time reduction just in drafting. And the quality often improves because you are spending your time on insight and not formatting.
00:10:26 Markus Kopko : Okay, let me show you an example. So sorry, that was a little bit too fast. So, here’s a concrete example. I will show you in a second short demo. So, first step is the input. In this example, we have a Jira export data, which is anonymised for this demo, or let’s say it’s synthetic data. Step 2 is to utilise an AI tool, which can be ChatGPT, Gemini, Claude, or maybe a proprietary AI tool you already use in your organisation; you name it. And step three is we will get an output and PRINCE2, in this case, a PRINCE2-compliant highlight report ready to review. And time needed, maybe about two minutes. Okay, so let me quickly change my window for the short demo here, just a second. I need to switch the window, so sorry for that.
00:12:25 Markus Kopko : This is demo 2 already, so I have to find the other one. Okay, here we are, sorry. As you can see, there is already an input file. This is the Jira data I talked about, and I have already put in a structured prompt here, as you can see, a prompt for generating a PRINCE2 highlight report. I’m not sure if you already confirm with prompting or prompt engineering, but this is not the topic for today, but a prompt should always have things like what is the context or the role the tool should perform, what is the input data, in this case the attached file, what exactly is the task, or what should the AI tool do exactly, and how should the output look like, in this case we want the sections for the PRINCE2 highlight report and so on. It’s, as you can see, a little already very comprehensive prompt here that maybe could be improved, of course, or customised to your specific needs. But yeah, this is just an example, okay? If you are wondering, I’m using here a tool it’s called Genspark. It’s one of my favorite AI tools, but you also can try this like with ChatGPT. I already put the file here too, and we can do a little experiment already.
00:14:32 Markus Kopko : I put the prompt also here and in Gemini, and let’s perform all the three tools, the same tasks. This is, by the way, something I do very regular, utilising not only one AI tool, but doing the same thing with two or three AI tools to just to compare the results because not any AI tool provides the same results for any task. So, you have this is a recommendation I can do from my side to find the tool which fits your preferences and your needs best and then stick to it. If not, your organisation already has an enterprise approved AI tool in place and you are demanded to use this one, but that’s of course another situation then. Okay, so let’s have a quick look on the results. As you can see, Genspark delivered a very comprehensive result here, we get a status summary, we get the reporting period, what everything is completed based on the data we have provided, what is still a work in progress. We get information about the status, also about the schedule status, what are the key issues right now. So, a lot of information, a lot of data provided to us in an already very structured format, right? Of course. In this case, we still have to put this in your standard PowerPoint report, or whatever format is needed. But of course, you can also utilise AI and create agents who even do this part of the task automatically in future, which just saves even more time. And to have a quick look in comparison, oh, ChatGPT is still working, as we can see, and Gemini here is already done and provides, I would say, a very similar result, but like I said, you can compare and then find out what the preferences are. As we can see here, the results from Gemini are not that comprehensive than the result from Genspark and ChatGPT is still working on it. Maybe I have activated extended thinking, so it’s really going deep into it and does some reasoning there, and so on. So, this take a little bit longer but yeah, I think you got the idea.
00:17:54 Markus Kopko : So much for now, let me see quickly switch back to the presentation. OK, yeah, we should see the presentation again right now. So, far for the first demo, quick safety note, but important safety note here, this demo, like I said, uses anonymised or synthetic data. You always have to follow your company’s AI governance policy, hopefully there is one in place already, and if so, you have to follow it before using any kind of data with, especially with external AI tools, right? But the principle holds; AI handles the compilation, as you’ve just seen, and you handle the insight, and that’s the shift we are making. So, second big win, meeting intelligence, maybe you already try that out. So, here’s the challenge: you are the project manager and most often, that’s at least my experience. You are not only the project manager but also the designated note taker in project meetings. That means 30 to 60 minutes of your attention is on transcription instead of facilitation. The AI solution, well AI transcription tools like Otter.ai or most often, maybe Microsoft Teams Copilot with the transcription feature already integrated, or Google Meet with LLM extraction. So, but what do those tools give you after such a meeting? Well, it locks the decisions made throughout the meeting. It can also provide action items even with owners if discussed throughout the meeting. The risks mentioned throughout the meetings and dependencies identified. So, a lot of things just provided right after the meeting in a structured way instead of creating the meeting minutes in a very long and tedious way afterwards and find out all the things by your own manually.
00:20:44 Markus Kopko : So, your role becomes this: you review just for accuracy, you add context, and you distribute the summary, time saved? Well, maybe 30 minutes per meeting and you are now facilitating and notetaking. Governance note here, check, once again, check your organisation’s AI policy, especially policy on AI transcription, and especially for confidential meetings. There might be a difference, but for most internal project meetings, this can be a game-changer. So let me connect these frameworks you already know and already use, maybe. So, if you’re using PRINCE2, for instance, AI as we have seen, AI automates highlight reports, also risk register updates, and you focus on reviewing outputs and managing by exception. If you’re using more PMI, PMP, AI handles, maybe EVM calculations or schedule updates, and you focus on resource negotiation and stakeholder alignment. And if you’re using more AgilePM well, AI tracks velocity and burn down charts, and you can focus on facilitating retrospectives and team coaching. However, key insight here is AI handles data compilation, and you handle human dynamics and strategic judgment, and that’s true regardless of your underlying framework.
00:22:46 Markus Kopko : Sometimes it works and sometimes not. So, before we go further, we need to talk about AI safety protocol, and this is non-negotiable. So, layer one is data privacy filter. Well, public data is safe for any AI tool, ChatGPT, Claude, Gemini, whatever, you name it. That can be examples here are industry benchmarks, open-source code, training examples, literally any information that is publicly available, okay. Internal data, it’s different, or this should only be used with enterprise-approved AI, like Microsoft Copilot 365 or Google Gemini Enterprise or ChatGPT Enterprise or whatever AI tool your organisation has approved. And examples here can be driver data or any data of your project management tools like project plans, even internal wikis, knowledge databases, that kind of stuff. Why is this important? Because this is not the topic of our session today, but shadow AI is a real problem nowadays. Many employees using AI ungoverned, uncontrolled, and this is, like I said, this is really a problem. Okay, confidential data, of course, should be human only or at least strictly controlled AI with a defined and documented audit trail. Examples here are like client contracts, for instance, employee reviews, financial forecasts, that kind of stuff. And when in doubt, then ask your IT or your legal team and don’t guess, or any instance in your organisation which is accountable for the AI governance. The layer two of our safety, AI safety protocol, is output verification. I think I mentioned that already. Always, and I really mean always, cross-check AI output against those data and also against your expertise. You are the expert, and AI makes failures and hallucinates; it has become better, meanwhile the model types have become better. There was much more hallucination in the beginning when, if you remember when ChatGPT was released in the beginning, it has become better, but it happens still, and it happens still often that AI hallucinates. So, you always have to cross-check the output because you are accountable for accuracy, and you should document your review process. And layer 3 here is traceability and governance. So, if possible, lock everything what you do in that context. Lock the prompt you used, the tool you used, the date, and who reviewed the results, if not yourself.
00:26:39 Markus Kopko : And once again, if one is in place, follow your organisation’s AI governance policy. And if there is not an AI governance policy in place, then initiate the conversation about creating one, because this is really important. And here’s why this matters: responsible AI use builds trust with leadership and trust unlocks budget. If you want AI tools approved, for your teams, show leadership that you are using them responsibly. Okay, act 2, the amplification phase. This is where AI doesn’t just save time; it makes you smarter. So, let’s talk about risk management. Traditional risk management looks like this: manual brainstorming, limited by your past experience, infrequent updates, and you probably misses subtle risks. AI-augmented risk management looks different; AI analyses waste data sets. Once again, coming from your systems, maybe Jira, meeting minutes, any kind of communication data. And AI identifies out of this waste amount of data emerging risks and delivers early warnings. “Similar projects often experience scope creep around Month 4 due to unclear requirements”. That can be an example inside of your risk management utilisation, AI utilisation for risk management and your decision, preemptively schedule a requirements workshop. So, AI doesn’t replace your judgment; it augments it, and it shows you patterns you wouldn’t see on your own. So, here’s another example I can show, once again, we have an input file with a Jira backlog + anonymised meeting minutes. And the simple prompt here would be analysed for hidden risks using PMP risk categories for the PMI people, but you can define, of course, which risk framework or risk categories should be used. And yeah, once again, you have to decide which tool you want to use, or if the organisation has one in place, then it should be used, this one, of course. So once again, I will do a quick switch to my sample.
00:30:29 Markus Kopko : Okay, once again, here this time, it’s another input file, of course. Like I said, it’s kind of a Jira backlog and also some meeting minutes, and here you can see the, just a second. Okay, this is the demo input here, what I have put in here, and just a second. This should have been the prompt, but that was my mistake. So, I replaced it because this is the data out of the attachment, but here is what it should have been: the prompt for the risk discovery here. So once again, this is a very comprehensive, structured prompt with providing context; the input data is the attachment, like I said. The task it should perform is analyse that input data and use the PMP risk categories, which are technical, resource, schedule, cost, external. Then identify risks that may not be explicitly documented but can be inferred from the patterns in the data. Then once again, here’s the risk framework it should use, and what our analyse requirements are and what the output should look like, and some additional instructions. This is, of course, my recommendation here regarding that kind of prompts, because they can become very sophisticated over time. You should create a kind of a prompt library within your organisation, or at least within your department. There are also tools available for doing so, I know about a Prompt Buddy tool for Microsoft Copilot, which works very well from my experience. And this way you will also be able to share your prompts with your team, with your colleagues, you can refine it and iterate it, and develop it further over time, that’s a good practice to do so. So, as you can see, Genspark was the fastest one again here. And it already discovered some risks, and not only discovered it, categorised it, delivers a risk statement, it delivers the evidence, so on which data realise that risk assessment here. It provides an assessment already regarding impact and probability, provides us already a score, and even some mitigation recommendations.
00:33:52 Markus Kopko : And this is for the, I think, top five risks we have defined here exactly. So, this was pretty fast, as you can seen and if you imagine how long you usually need for this kind of work, creating a risk register with all that information throughout workshops or brainstorming meetings with your team, then you get a pretty good impression on how big the time savings can be by utilising these kinds of tools and these processes. OK, I will jump back to the presentation. So, here on the slide, you see just three examples, for examples for discovered risks as about different categories like technical risk, resource risk, and schedule risk. And as you have just seen in the tool, it even discovered some more risks. And of course, you can, once again, you can customise this to your needs and your preferences. If you say, okay, I will only always see only the top three risks or better the top ten, that’s of course up to you. But as you have seen, you get a comprehensive result here, but your role is still, of course, you review the results. Like I said before, always cross-check, you validate it, you act, you decide what the next tasks are, or the next steps are, and that’s amplification. So, AI surfaces the insight, but you make the call, and another amplification use case can be plan review. The problem after hours of planning, it’s very easy to miss critical gaps or assumptions or just becoming unconcentrated and doing failures. And the AI solution here, well, ask AI to review your plan, for example, for missing dependencies, acceptance criteria, maybe unrealistic estimates, and resource conflicts, all that kind of stuff. And the benefit is you can catch planning errors before kickoff and not during execution. So, this is like having a second pair of eyes on your plan, someone who’s not tired, not biased, and has seen thousands of similar plans before. If you are able to, if you have a historical project database in your organisation and you are able to utilise all this historical data for such a prompt or such an AI agent, maybe, this is like a gold mine, you know.
00:37:32 Markus Kopko : So now I want you to show a high-risk use case. It’s important about to talk about all those things. So, this use case requires legal and human resource approval before implementation, but you’ll also be able to do so if you want. So, AI can analyse metadata, not just content, but metadata from your project communications, like response rate trends, or how quickly and consistently stakeholders respond. Meeting participation patterns, attendance and engagement levels, milestone review attendance, all those things, monitoring presence at critical sessions. And an example signal here could be that stakeholders C’s response rate dropped from 90% to 30% over the last two sprints. They missed the last 2 milestone reviews, and you can utilise this insight and take action. Maybe you go into a proactive one-on-one meeting to address potential disengagement. This is powerful, but it comes with risks. So let me be very clear about the warnings here. Three critical warnings: do not analyse personal communication without explicit consent. Do not analyse tone or sentiment of individuals, that’s a privacy violation risk. This always requires approval from legal and HR before implementation. Safer alternative, of course, manual tracking of objective metrics, response rates, attendance, things you can see without AI. So, when in doubt, don’t do it. Relationship management beats surveillance always.
00:40:02 Markus Kopko : So let me be clear about the role of AI in decision making, AI is always decision support or should always just be decision support, not decision maker. So, for PMP risk management, as we have seen, AI provides pattern recognition from historical data, but you provide the contextual judgment. For instance, “AI says low risk, but I know, or you know this vendor is unreliable”, you override. Or for PRINCE2 business case, for instance, AI provides real-time ROI tracking and scenario modeling; you provide the strategic narrative. “Numbers say stop, but market positioning says pivot, you override. Or for AgilePM sprint planning, AI provides velocity forecasting based on team history; you provide the team dynamics read. And maybe “AI suggests 15 points, but team is burned out”, you adjust. So once again, key principle, AI provides the data. You provide the context and both together equals sound decision. Let me show you the real-world impact here. So before, traditional project management workflow, high admin burden, reactive risk management, if there was even some risk management, delayed issue detection, limited strategic focus. After, AI augmented workflow, reduced admin by three to five hours per week, proactive risk mitigation, more strategic time, enhanced team and stakeholder focus. Results, well, three times faster project completion, maybe, 85% team engagement, 92% stakeholder satisfaction. So, these patterns are observed in many integration projects I was involved. So, this is real and it’s happening right now. So, act 3, the leadership phase, this is where you focus on what I cannot do, and here’s what I cannot do. Complex negotiation, navigate ego, politics, and power dynamics. Example, resolving sponsor conflict when AI data is ambitious. Ethical oversight, audit AI for bias, ensure fairness and transparency.
00:43:27 Markus Kopko : So, example here could be overriding AI resource allocation that maybe discriminates. Visionary alignment, keep project aligned with strategic “Why” despite data signals. Example here, pivoting when market shifts, even if metrics are green. Or change leadership, lead teams through AI adoption and transformation. Building psychological safety during AI integration. If you master this, AI becomes your unfair advantage. So let me show you a real override moment. AI recommendation was “Cut Team B, they have 40% lower velocity than Team A. Reallocate to higher-performing team”. Logic, data-driven, and efficiency optimised. Project management override, your override Team B is onboarding 3 junior developers; velocity dip is temporary and expected. Decision, reject AI recommendation. Maintain team B investment. Or leadership principle, “AI optimises for efficiency, but you optimise for long-term value and people development”. This is the human critical zone, and this is where you are irreplaceable. Another override moment, AI suggestion, cut “low priority” accessibility features (used by 5% of users). Logic? Well, focus on schedule optimisation, low usage percentage, high development cost. Why should you override, or why could it be override? Well, cutting accessibility violates company values and regularity requirements. So, ADA or EU AI Act compliance, maybe for instance. So impacts the critical 5% of users, then the decision should be extend deadline to launch with full accessibility. Outcome here is product integrity maintained, brand protected, legal compliance ensured. So, this should just show you how important that human oversight, that human in the loop moment, really is. Leadership principle, “Data informs decisions. Values guide decisions”. So, we covered 3 phases: liberation, amplification, and leadership. We’ve talked about a few tools, AI governance, ethics, and also real-world impact. Now I’d like to open the floor for your questions. So what challenges are you facing? What use cases are you curious about? Or maybe what’s holding you back?
00:47:12 Sevcan Valiyeva : We do have a few questions in the chat, but just before we head over to the Q&A, I would like to remind everyone that this webinar is being recorded. I will send you all the recording and also the presentation slides that I believe one person specifically requested. Markus, can you go on to the next slide, please?
00:47:37 Markus Kopko : Yeah, just a second.
00:47:46 Sevcan Valiyeva : The next one.
00:47:47 Markus Kopko : Next one, okay.
00:47:49 Sevcan Valiyeva : So just very briefly, these are all our courses. If you are interested in any, please let me know. I will give you more information, specifically for this webinar, the most appropriate would be the Project Management courses, PRINCE2 and AgilePM. I’m going to write my e-mail here. So, if you have any questions, you can always e-mail me. I’ve also got a very short survey, I’m going to put up, just while people answer, I’ve got a question. Yeah, so the first question we have for Markus is whether he could provide the AI prompts that he used.
00:48:46 Markus Kopko : Yeah, sure, no problem. I will send it over to you, Sevcan, and maybe you can then provide it to the participants.
00:48:53 Sevcan Valiyeva : Yeah, that’s great. Okay.
00:48:55 Sevcan Valiyeva : We’ve got another prompt question. So, it’s, are there any go-to daily prompt he uses for daily PM work? If so, is it possible to share them with us?
00:49:10 Markus Kopko : I’m not 100% sure if I got that correct. So, what is meant by PM daily ramp-up in that case?
00:49:19 Sevcan Valiyeva : Hassan, if you can just clarify that, but my understanding is maybe if you are using one specific prompt for daily, use daily, I guess. If not, we can always wait for Hassan to reply.
00:49:38 Markus Kopko : Yeah, yeah.
00:49:42 Sevcan Valiyeva : The last question is, there an AI GPT that will log for you, i.e. The prompt used, tour used, date, and reviewer.
00:49:58 Markus Kopko : No, not a GPT, or when I got that correct, something like an agent or something like that. This is something which falls under AI governance aspects, and if an organisation decides to use any kind of AI tool, if it is something like Microsoft Copilot, for instance. Of course, Copilot comes with some such kind of logging, and it’s also configurable to some extent, I think. But my answer would be if you or your organisation is serious about implementation, such kind of AI empowered processes or maybe even AI agents in the future, then this kind of governance and audit should be incorporated into that process. Then you will be, this is a bit of development that is then necessary, but if it is built in that process, then that would be the best solution. But I’m not aware about any, let’s say, custom GPT or something like that, which will do specifically this part.
00:51:29 Sevcan Valiyeva : Thank you. I hope that answers your question. So, Hassan replied, he’s asking for any prompts he uses, say, post-meeting summary or reporting, so on and so forth.
00:51:44 Markus Kopko : And if the question goes to me, so if I’m using such kind of prompts every day, was that the question? Sorry.
00:51:51 Sevcan Valiyeva : Yeah.
00:51:53 Markus Kopko : Well, to be honest, I have not a standard prompt which I’m really using every day and every time in the same or similar way. It’s more like, situation-specific. What I have done in the past for my past employer or also for customer organisations is building, like I mentioned before, building a kind of a prompt library to support their standard project management processes, like, for instance, like creating a project charter. Of course, this is a task which repeats regularly every time a new project is initiated; there should be a project charter, right? And to support the project management organisation in creating that project charter, in a standardised, and professional, and repeatable way, we created a prompt, a generic standard prompt for their organisation, but customised, of course, to their needs and their requirements. So, if you search the internet for such kind of standard prompts, you will find a lot, you will find hundreds, if not thousands, of examples. Even just for creating a project charter, if you Google or ask ChatGPT for providing such kind of prompts, you will get a lot of response. But they are always very simple, you know, and very, very, not that comprehensive, like the example prompts I have shown you earlier. So, the reason behind that is that you always need to tailor those prompts to your specific organisation environment, to your specific project management process requirements. And that is what makes those prompts individual for any organisation. But within the organisation, it can be used in a generalised and standardised way. I hope I made the point clear, I’m not sure. Sorry english is not my mother language, so maybe that’s the reason.
00:54:30 Sevcan Valiyeva : He did actually say good shout on creating a prompt library for repeatable PM works. So, I am hoping you did answer that question. We’ve got another question: how do AI algorithms predict project success based on historical data?
00:54:50 Markus Kopko : That’s a question that need to be asked. I would say in AI expert in terms of algorithms, in terms of being a developer. I’m not such a developer, and I’m not that deep into the technical aspects and into the algorithms, but what I can say is. You always need to define the problem you want to solve with AI, regardless if it is in that example.
00:55:35 Sevcan Valiyeva : Before the webinar, he did say his internet was quite bad. Let me see, sorry about that, everyone. I think it’s back, I heard a sound, maybe not, let’s just give a few minutes.
00:56:25 Sevcan Valiyeva : Okay, let me end the survey while he enters. Okay, you are all free to leave, I don’t want you to be waiting for Markus. I am thinking he has internet problems, which I’m sure we all have. Okay, he just moved the slide. This, even this section, will be recorded, so you can always catch on if Markus manages to join. I think there was one more question that we had from Raza Hussain. I can message you individually, specifically with that answer if, you’d like, because obviously I don’t want you to be missing out on anything. I hope that’s okay. I’m just going to wait a few more minutes, just in case he manages to join. Once again, I do apologise.
00:58:27 Sevcan Valiyeva : James, I will e-mail you with more information about that if that’s okay. I’m just going to write your name down. You can try that, Markus, yeah, that might be useful.
00:59:21 Sevcan Valiyeva : It does seem like Raza’s question is popular. So, I can include that in the e-mail if people would like to leave, and you can always catch the answer in the e-mail or again, the recording. So, Markus has joined. Let’s try again.
00:59:47 Sevcan Valiyeva : Yeah, we’ve got you.
00:59:50 Markus Kopko : I’m very sorry, that was very strange. All the time, I can see and hear you, but you may not.
01:00:00 Sevcan Valiyeva : Yeah, that was very strange, but it doesn’t really matter. I’m sure it was just an internet problem. Welcome back. So, we’ve got a question, let me just go up so I can find the question.
01:00:15 Markus Kopko : So, I think the last question was about how AI can predict project success based on the historical project data. And like I said, I’m not sure what was the last, what you heard from me, but I said this is a question that need to be answered by someone who’s really deep into AI development and algorithms, which I’m not, I’m more on the managing side. However, those kind of questions are not easy to answer because how does that organisation defines project success? This is the very first question you need to answer here because from my experience, any organisation defines that in different ways. So, the general answer once again here is you need always. The very first step is you need to define very clearly what is the problem you’re trying to solve. And if you have defined that problem, you should be able to matter it, to connect it to one of the so-called 7 AI patterns. I’m not sure if you’re aware about that term; this is coming from the PMIC, PMAI framework, which I’m a big fan, by the way. And if you are not able to connect your problem, let’s say your business problem, you are trying to solve to one of those 7 AI patterns, which is for just a quick example, one of those patterns is pattern recognition, as we have heard before. And if you are not able to connect that to at least one of those patterns, this is a very strong signal that AI might be not the right path to follow to solve those problems. If you are able to connect it to at least one or maybe even more of those patterns, that in the opposite is a strong signal that AI might be the right solution. And then the next step is to think about the right model. You should, you do need for that specific problem, and the integrated in the model, the algorithms, the underlying algorithms that should be applied to solve that problem. So, that can be a very complex situation and very complex questions to be answered. But the fact is, AI can help to predict that probability of project success. Especially if you have a large database of historical data, as more data and as better the algorithms and the model, as more accurate the predictions can be, of course.
01:03:24 Sevcan Valiyeva : We’ve got a final question. Any recommendations on what to use from the AI outputs to a graphical system for board members’ easier reading?
01:03:40 Markus Kopko : Okay, I think there are specific, very specific AI tools out there. I showed you one before, which was one of my favorite tools. It’s like I said, it’s Genspark and Genspark is more a collection of AI agents. And one of the agents is a document agent, which is really able to create documents from those outputs. So in the general chat you have seen before, you put in your prompt, you get your result, and if you are satisfied with the result, you can say, okay, now please create me a Word document, a PDF, or even a PowerPoint. It has a very, very powerful slide agent integrated, which is really good in creating PowerPoint slides, even based on your CI template. But that’s a very specific tool, and you know, it also does cost something. And I’m not sure if an organisation, an enterprise organisation, is willing to utilise that. But yeah, I have experimented with MS Copilot in the very early time when a Copilot was rolled out. And I was very disappointed about the capabilities in creating and managing PowerPoint slides. I’m not sure if this has improved significantly over time right now. I’m pretty sure it has improved, but I have not tried out recently. So yeah, maybe MS Copilot can do a good job here also in creating Word documents and that kind of stuff. But yeah, this can be a problem, I understand that question.
01:05:57 Sevcan Valiyeva : If anyone doesn’t have any more questions, obviously feel free to write them in the chat. I would like to say a big thanks to Markus for joining us, really appreciate it. I hope all of you enjoyed the webinar and learned something. If no one has any more questions, I am going to wait for a minute or so, just in case anyone has any questions, but you are free to leave.
01:06:25 Markus Kopko : Sevcan, thank you for having me. It was a pleasure, and thank you all for your time and for listening. And if you like, feel free to reach out at LinkedIn, for instance, to connect. Thank you very much.