Project Management And AI: What Should Be Automated And What Should Not

By Jason C. Bork, president and founder, Pintail Solutions

I was working in life sciences project management at a large pharma company when a new project manager I was mentoring came into my office after only a few months on the job and said, “I don’t know that we need to meet anymore. I’ve got it now.”

I was confused. He had come from Discovery Chemistry and had very little experience seeing how multifunctional work integrates across chemistry, biology, regulatory, clinical, drug disposition, legal, commercial, quality, CM&C, and other functions. While the concepts of project management may be applied in some form across most jobs, successful project management in life sciences is far more nuanced. It requires navigating Ph.D.s, MDs, functional leaders, executive stakeholders, and governance boards, as well as scientific uncertainty, budgets, timelines, and risk.

So, I asked him what he meant.

He explained that his timeline, budget, and risk plan were all pulled together. In his mind, he was in good shape. I asked what he was doing with those tools. He wasn’t sure what I meant. After some back and forth, he said they were all on SharePoint and therefore “done.”

I hope you see the problem.

Project management is not really about the tools. I may have once said, and was told never to repeat, that project management tools could be completed by non-human primates. But successful project management requires deep judgment. The tools provide data. The real value comes from using that data to influence decisions, often with individuals multiple pay grades above you or with executive governance committees responsible for major portfolio choices.

In a life science data-driven ecosystem, that is where the power, influence, and necessity of project management as a function truly resides. And AI is making project management even more powerful when it is used correctly.

For this article, we will use one of the simplest and most durable project management methodologies popularized by W. Edwards Deming: Plan, Do, Check, Act (PDCA). Through that lens, we can explore what should be optimized by AI and what should not.

The answer is not “automate everything possible.” Life sciences projects operate in highly regulated environments where patient safety, scientific integrity, regulatory compliance, and strategic decision-making are paramount. Successful organizations will use AI not to replace project managers but to augment them. AI should automate repetitive and data-intensive activities while preserving human oversight where judgment, ethics, accountability, and stakeholder alignment are essential. AI can also apply learnings across multiple projects in ways that would be difficult, or even impossible, without AI-assisted integration.

Plan

Identify an opportunity, define the objective, and establish success metrics. Assess the current state and build a strategy or hypothesis to enact a change.

AI can recommend. People decide.

For organizations with sizeable portfolios, AI should significantly improve planning. It can integrate prior planning and execution experience to develop an initial draft plan for team review, discussion, and adjustment. AI can draw from a broader set of previous project experiences than any individual project or program manager could reasonably hold in memory.

It can also compare baseline plans against actual durations experienced during execution, which should lead to more realistic planning. When past durations and costs are incorporated, future risk and budget planning can also improve. Organizations with fully integrated planning and execution systems can use AI to forecast and optimize resource constraints across functions. The potential value of that intelligence is significant.

However, it also creates risk.

I have seen planning templates used as starting points for projects. Without fail, those templates are often reviewed and accepted as “good enough” without meaningful evaluation by the team. If team members do not fully consider what is unique about a project, where the project sits within the portfolio, how much parallel processing or risk is appropriate, and what should be done differently given the project characteristics, then the planning process has failed.

A project with a high regulatory hurdle, low toxicology risk, or a competitive race to first-to-market should not be planned the same way as every other project. Templates and AI-assisted planning can make it easier for team members to shed accountability with statements like, “I never agreed to those timelines.”

The phrase “human in the loop” is used constantly in AI discussions, and it is absolutely crucial here. Team members must apply critical review and judgment while remaining open to learning from other projects through AI. That is how planning becomes both more informed and more accountable.

While at a large pharma company, I led the implementation of critical chain project management following the concepts defined in the theory of constraints. We were able to aggressively accelerate high-priority projects. The methodology truly allowed us to think differently about each project and remove operational hurdles. However, it really scared people.  We could treat some projects differently, but we could not deliver the entire portfolio of projects in the same accelerated way. Teams were concerned that every project would be expected to deliver against a hyper-fast timeline, and it wasn’t possible either internally or with external partners. AI can help develop accelerated plans and some of these ideas might be transferable across all projects, but some will not. Maintaining judgement, in part, based on the project’s portfolio prioritization is critical and communicating the assumptions and risks to stakeholders and governance committees is key.

Do

Implement the plan on a small-scale or pilot basis. Execute the planned actions, gather necessary data, and document any issues encountered during this test phase.

AI manages processes. People lead people.

There are real benefits for AI in this phase, especially on the tactical operations side of the business. Generating draft agendas based on timelines, risks, budgets, and priorities is a great use case. So are meeting minutes, action item capture, budget monitoring, resource allocation tracking, routine reminders, and standard communications.

But execution is where conversation and connection become critical.

The conversation cannot start and end with, “Are you going to hit your end date?” What happens in the middle matters most.

When I was a very young and perhaps naïve project manager, my program team leader stopped by my desk to check in. I assured him that I had sent emails to all team members asking for timeline updates and potential risks. He shook his head and said, “Follow me.”

We walked the halls to visit team members in their offices and labs. The amount of information we gathered was exponentially greater than what I would have received by email.

For example, my drug disposition scientist assured me we would hit our target date for data delivery. I wanted to get up and leave, but something told me not to. I also knew he was notoriously late with deliveries and always extremely busy, so I started asking follow-up questions.

What still needs to be done to get the data?
How long is the queue to get the study scheduled?
Are we in the queue yet?
Are we actually scheduled?

That is when we both realized we were already late. We had to execute a workaround to have the data available on time.

In addition to the technical project management details we uncovered, we also strengthened our relationship. That matters. Relationships are critical to building commitment, resolving conflict, and negotiating priorities. People must manage vendor and stakeholder relationships. Teams must make scientific and technical decisions. Leaders must adapt plans when circumstances change.

AI can support execution. It cannot replace the human leadership required to move complex work forward.

Check

Review the pilot test and compare actual results against expected outcomes. Analyze data to understand what worked, what failed, and identify remaining gaps.

AI provides visibility. People provide insight.

AI should drive operational efficiency and improve visibility to organizational trends. It can monitor KPIs in real time, generate dashboards and executive reports, and summarize project health, emerging issues, and risks.

With a fully integrated system, AI can also analyze clinical, manufacturing, or operational performance trends not only within a single project but across the portfolio and across functions. It can help predict capacity constraints and identify consistent performance patterns that leadership may need to address at an enterprise level.

I love the continuum of data, information, knowledge, and wisdom. AI should optimize the data and provide new information that would be difficult to identify without it. But people must provide the knowledge and wisdom.

That means interpreting findings with scientific and business context. It means evaluating the significance of risks and determining escalation priorities. It means assessing patient, regulatory, and commercial implications. People provide the insight that turns visibility into action.

Act

Take action based on what was learned in the Check phase. If the test does not succeed, revise the plan and restart the cycle. If succeeds, standardize the change, implement it more broadly, and repeat the cycle to drive further improvement.

AI recommends actions. People are accountable for outcomes.

AI can help consolidate lessons learned across a portfolio of projects. It can recommend process improvements, draft dos and don’ts, develop top 10 lists, and document learnings not just from one project but across the broader portfolio.

This is one of the most commonly missed steps in portfolio management. Teams often celebrate and move on to the next milestone or they disband and take on new work. Capturing lessons learned from a single project is useful. Combining those lessons across a portfolio creates the opportunity for deeper learning and broader operational impact.

But people must continue to bring judgment and decision-making to this phase. People provide interpretation, context, and perspective. People communicate changes and ensure they are applied to future projects. That discipline can positively affect compliance with regulatory and quality requirements.

Most importantly, people remain accountable for the results.

AI As A Multiplier, Not A Replacement

In life sciences project management, AI should be viewed as a multiplier for the PDCA cycle. It can dramatically improve the speed, accuracy, and consistency of planning, execution, monitoring, and continuous improvement.

But accountability must remain with project leaders, project managers, and governance bodies — not only because life sciences projects affect patients, regulators, and public health but because each project is unique and must be managed with appropriate judgment.

AI can organize the data.
AI can accelerate the analysis.
AI can recommend options.
AI can surface patterns that humans may miss.

But people must decide what matters, what is ethical, what is compliant, what is strategically sound, and what will actually move the project forward.

Project management has never been about completing the tools. It has always been about using information, influence, judgment, and leadership to deliver meaningful outcomes. AI does not change that. It simply raises the bar for how well it can be done.

About The Author:

Jason C. Bork is a life sciences executive with more than 30 years of experience across large pharma, CRO, and startup organizations. He is an entrepreneur known for seeing the essence, distilling the complex into actionable steps, and developing those around him to new levels of fulfillment. From business strategy, organizational change, and operational excellence, Jason enables organizations to solve critical challenges, deliver high-impact projects, and move into the future with clarity and confidence. He has authored multiple scientific papers and book chapters and is a distinguished public speaker. Jason founded Pintail Solutions, a life sciences consulting organization, in 2015.