AI and Project & Portfolio Management – Red Herring or Golden Opportunity? – Part 2

Could AI offer novel and impactful solutions for some of the complex challenges being faced in PPM?

Part 2

?

Jamie Barber

?

In Part 1 ((17) AI and Project & Portfolio Management – Red Herring or Golden Opportunity? - Part 1 | LinkedIn) I looked at how AI could be applied in the field of Project and Portfolio Management (PPM) as a time and effort saver, freeing up project resource for the more value-add activities that humans do best – activities that can help mitigate some of the key causes of project failure. In Part 2 I’ll be considering how AI could be engaged as more of a strategic enabler – to ultimately adopting more of a ‘trusted project assistant’ role.

?

AI as Researcher and Data Analyst

Now we start to dive into those PPM activities that could benefit from the hugely powerful number-crunching capabilities of AI, especially when trained on your organisational data in addition to the wider internet. Projects are data-generating behemoths and this data is complex, potentially unstructured and part of a rich interwoven tapestry, not only within the project but also as part of your wider organisation and external world. This data contains clues and patterns which may be self-evident or may be buried deep in the detail. Sometimes it will only be by examining hundreds or even thousands (or more!) data items and looking at how they interrelate that will uncover a pattern of predictability or gem of actionable insight.

Enter AI. It can juggle data items and dependencies on a mind-bogglingly large scale, certainly more than a human could match. And while there are challenges (not least in needing to consider historical bias and incorrect or incomplete data), its power can be deployed to enhance decision-making and supplement the work of the Project Manager.

You can already access any reputable Generative AI tool today (ChatGPT, Copilot Bing search as examples) and ask for suggestions of risks which may impact your type of project, and even suggested mitigations. The results returned will only be as good as the context you provide in your prompt, and you’ll need to do some weeding out. But take this one step further and imagine if you were able to call not only on industry best practice but also on data captured for every project historically delivered in your organisation that left a data footprint - regardless of geographical location or outcome.

You are now in a world where you can generate initial drafts of whichever project artefact you choose, from Lessons Learned to Non-Functional Requirements lists and Stakeholder Matrices. What are the key learnings of past projects you should be aware of? Which grades of seniority need to be involved in approvals and sign offs? Which vendors were used and with what outcomes? Which checklist items are needed to ensure a robust service acceptance into BAU? Yes, this data may exist today but as a Project Manager would you really have time to wade through it all (for every historical project?!?) to find the proverbial needle in the haystack? A note of caution - when looking at historic data bear in mind that vendor capabilities, market conditions, and a myriad of other factors may have changed. I’m not advocating using this information in isolation – you’ll still need appropriate stakeholders (Project Manager, Business SMEs, Change Managers, Business Analysts, Architects) to add in their expert knowledge and appreciation of context. But how much time could be saved by having a ‘starter for ten’ based on what your organisation has delivered in the past ready to amend and adapt as required? The link to your historic organisational data is the key, and that may be the next avenue of exploration to fully unearth the next raft of benefits of AI.

?

AI as First Responder

Project teams are inherently fluid and change as team members roll off or new people join. By training AI on a dataset that includes not only data about your current project but also wider information from HR, Compliance, or Facilities functions, the new starter could have a ready-made ‘buddy’ to accompany them during their initial onboarding – one which they can pose questions to in simple language. So the Project Manager can focus on whether the new team member is happy, engaged, and motivated, whether they are using their skills in the way envisaged, or how they can advance their career aspirations on the project. Rather than providing updates on day-to-day progress, how to submit expenses, or where the toilets are.

The Natural Language Processing (NLP) capabilities offered by AI - even when referencing unstructured data sources such as reports, emails, and chats – make a further compelling case for its use as ‘First Responder’ both to queries arising from inside and outside your project team. Enhancing dashboards with this wider, richer, real-time dataset means that interested parties can easily dig further. Imagine how much more quickly a stakeholder could ask questions in plain language and perform an initial ‘Five Whys’ query on reasons for a particular project delay (see Use the 5 Whys Method for Better Problem Solving (liquidplanner.com)). Or ask which upcoming risks will have the greatest impact on delivered outcomes. All in simple language. The stakeholder can then have an informed discussion with the Project Manager about impacts.

It could be argued – whisper it – that the days of Project Managers collating data and producing dashboards and reports are numbered. The PM becomes instead the person who helps the stakeholder interpret what story the data is telling in response to a plain language question, and provide expert analysis, mitigations, and corrective action recommendations rather than crunching numbers.

?

AI as Strategic Advisor

Again, we are into a ‘horizon view’ here, but imagine if you had trained your AI tool not just on data related to project delivery but also on operational data, usage patterns and integration information for systems impacted by your new project. Not only logs of users and actions performed but error messages encountered, time spent in various processes, workflows not completed. You now have the basis for a real-time As Is Process and Architecture Map that will almost certainly be a more accurate (and speedy) indicator of how the current estate is actually used (rather than a document stating how it should be used) and where the pain points and bottlenecks are hiding. This may help to inform your benefits case, as well as identifying As Is areas which will be negatively impacted and may incur increased technical debt (your disbenefits). It will also highlight potential dependencies to and from your project and other projects either in delivery or BAU.

?

A considerable strategic advantage which should not be overlooked is the deployment of AI in portfolio planning, sequencing, and prioritisation. What are your organisation’s current strategic objectives? Maximising revenue, decreasing costs, reducing technical debt, addressing regulatory requirements? More than likely these and more, with some combination of factors with appropriate weighting that changes over time and market conditions. Imagine all the data points involved in portfolio planning should you need to tweak the input variables.

?

You might need to ‘dial up’ the importance of reducing operational costs or ring-fence a larger proportion of budget for an audit requirement or introduce a new high priority demand. Perhaps a particular project undergoes a scope change during the project lifecycle and the portfolio then needs to be adapted to deliver in future more of the benefit that has been dropped. Perhaps a key person leaves the organisation. Maybe a foundational programme with many dependencies on it suffers a delay. You would need to refactor time, cost, resource allocation, and dependencies on/from other projects, not to mention the timeline of benefit delivery against your overall organisational strategic objectives. And you should be doing this refactoring every time an input variable changes. This is challenging and time-consuming even with an established portfolio management software tool with access to those data points, let alone if you are relying on a more basic manual tool for portfolio planning. Imagine having an AI assistant following the thread and updating all the downstream data items and impacts, and then returning promptly to let you know what this updated end state looks like.

?

As well as needing to turn on a dime when input factors actually change, the Portfolio Manager will derive huge value and be able to have option discussions with senior stakeholders if they are able to conduct ‘what if’ analyses. What if strategic priorities were updated? What would happen if a combination of risks morphed into issues? What if two projects were merged or two Project Managers with differing skillsets were swapped? What if all these things happened at the same time? AI can run simulations quickly through multiple different scenarios with different data variable combinations (and by ‘multiple’ I mean literally millions and more…) to identify the optimum scenario to work towards. And the input data can be changed, and the simulations rerun as needed. And the predictive capabilities of AI enabled by recognition of patterns in huge datasets can even be used to provide the most likely ‘what ifs’ that you should be looking at in more depth.

?

And as well as the Portfolio Manager deciding which input variables to change, the ‘what if’ analysis can be run with an output focus – what is the optimum combination of inputs to produce an outcome that is the most efficient utilisation of resources in a specified time period, or that is the cheapest, or that is delivered in the quickest time with the given trade-off of benefits. ‘What if’ analysis can be particularly valuable when data variables have a non-linear impact, making other ‘downstream’ variables change in unexpected ways and highlighting non-obvious linkages. And ‘what if’ becomes more attractive as AI capabilities grow and the organisational appetite for making better use of its own historic data increases.

?

Conclusion

While it will not be the silver bullet to magically address every concern, as outlined in Part 1 AI can play a huge role in removing some of the more administrative parts of a project professional’s day-to-day working life, enabling them to focus on the elements in their skillset that cannot currently be addressed by a non-human actor. Imagine the hours saved in administrative tasks that could instead be devoted to those real value-adding areas. Conflict resolution, negotiation, blocker removal, and team coaching and motivation to name a few.

But when we enter strategic territory, think for a moment of the step-change that could be realised. We could be using AI to link data items together and running scenario analyses with a volume of data that is currently just not feasible with traditional portfolio planning tools. We can be making sure that we are delivering the right projects – those that are aligned to strategic objectives – and having conversations with senior stakeholders about what this delivery will look like given risk profile, resource plan, timescales, dependencies, and benefits delivered. And tweaking those data items on the fly.

Project professionals will undoubtedly have a huge role to play, but they will be deploying their skills where they can truly add human value. In the meantime, a considered deployment of AI will help delivery evolve from the traditional project management triangle of outputs (time, budget, and quality) towards outcomes and greater alignment to strategic objectives. And provided we can adapt as project professionals to take advantage of the opportunities, then we are well-placed to reap the benefits, both within our projects and in our career advancement.

As Project Managers, we all love a good Red/Amber/Green assessment. Where would you RAG your organisation’s current and desired capabilities against the six roles of AI in PPM outlined in Parts 1 and 2 of this article?

I think the future is bright. Let me know your thoughts. Red herring or golden opportunity?

?

?

?

?

?

Jamie Barber is an independent researcher and thinker with extensive experience in Project and Portfolio Management, coupled with a keen interest in AI developments and their application to the project world. He can be contacted either via LinkedIn (Jamie Barber | LinkedIn) or [email protected].