AI Can Accelerate Value. It Can Also Accelerate Waste.

Why project fundamentals must come before automation

“A bad system will beat a good person every time.” W. Edwards Deming

“There is nothing so useless as doing efficiently that which should not be done at all.” Peter Drucker

AI is a remarkable accelerator. That is precisely why organizations should be careful where they point it.

In a sound project system, AI can shorten cycle times, improve analysis, and relieve teams of low-value administrative work. In a confused system, it can produce confusion more quickly, more consistently, and with formatting polished enough to make the project look like it is on track.

The Project Management Institute recently published an article examining how AI agents fit into a RACI matrix. Its central conclusion is sound: an AI agent may be Responsible for a bounded task, but it cannot be Accountable. A human must still review consequential work, approve decisions, and own the result.

That is the right answer. But it prompts a question one step earlier: Does the underlying system know what it is asking AI to accelerate?

The problem AI did not create

In working with hundreds of organizations of every size, across many industries, I have seen a recurring pattern. The fundamentals of project management can easily become casualties of urgency.

Projects begin before the business need is fully understood. Charters sometimes serve as permission to start rather than agreement about why. Vision statements are broad enough to support several interpretations. Success criteria are deferred until someone asks what success is supposed to look like, normally too late in the project life cycle. Requirements change while traceability isn't maintained.

None of this means people are careless or incapable. Usually, the opposite is true. Capable professionals are balancing customer expectations, executive demands, limited resources, shifting priorities, and deadlines established before the work revealed its full complexity. Under that pressure, organizations may substitute motion for alignment because motion is visible. Meetings happen. Deliverables appear. Dashboards acquire colors. The project is undeniably active.

But is the project delivering the right thing?

Then AI enters with the promise of speed and productivity. It did not create the missing clarity, but it will inherit it.

Lean asks the question busy organizations tends to skip

Lean does not oppose speed, it opposes waste. Its objective is the efficient flow of value to the customer.

Before automating work, Lean asks us to understand the process, identify what creates value, expose waste, reduce variation, and improve the system. The first question is not, “How can we do this faster?” It is, “Should this work be done this way at all?”

Automation should follow understanding. Otherwise, efficiency becomes the ability to perform the wrong work with impressive consistency. If a process already produces ambiguity, rework, delays, or outputs disconnected from business value, AI will not correct those conditions merely by operating faster. It will scale them.

Accelerating a bad system is not transformation. It is expedited waste.

Consider project status reporting, an obvious candidate for AI assistance.

An AI tool can pull schedule information, summarize risks, organize stakeholder comments, and produce a polished executive update in seconds. That may eliminate hours of administrative work. But what if the project never established meaningful success criteria? What if milestones measure completed activities rather than delivered outcomes? What if requirements cannot be traced to the business need? What if stakeholders are operating with different definitions of success?

The AI can still produce a concise, professional report. It can arrange the available information into a persuasive narrative because that is what it was asked to do. That can create false confidence. A report that appears organized and precise invites trust, and AI is exceptionally good at producing that appearance.

A human reviewer may catch an unsupported fact or a questionable conclusion. That reviewer cannot validate performance against success criteria that do not exist or verify traceability that was never maintained. Review is an important control. It is not a substitute for a sound system.

Yes, AI can help build the fundamentals

At this point, the reasonable AI advocate raises a fair objection. AI can help teams draft charters, develop vision statements, identify stakeholders, propose success measures, document requirements, and build traceability matrices.

Correct. It can, and we should use it for those purposes.

AI is excellent at accelerating the mechanics of project management. But generating an artifact is not the same as creating alignment. An AI-generated charter can be produced in twenty seconds and still be no more meaningful than a template completed in twenty minutes. The document is not the governance. The governance comes from the decisions, conversations, commitments, and accountability represented by it.

AI can recommend a course of action, but it cannot substitute for agreement among stakeholders. It can describe accountability, but it cannot make someone accept ownership. Use AI to prepare difficult conversations, identify missing information, and document agreements. Do not confuse the speed of document production with the quality of the agreement.

Foundational does not mean optional

The PMI article includes a statistic that should dominate more of the AI conversation. Only 37 percent of projects establish success criteria early, use a measurement system to guide decisions, and track performance throughout the project. PMI reports that project success scores doubled when all three conditions were present.

If three established management practices can produce that difference, the opportunity is not simply to teach AI where it belongs in the RACI. It is to practice sound governance before delegating work within it.

Before asking where AI belongs in the RACI, organizations should be able to answer a few basic questions:

1. What business need are we addressing?

2. What value must the project create, and for whom?

3. How will we know whether the intended outcome was achieved?

4. Who has authority to make which decisions?

5. Can requirements and deliverables be traced to business objectives?

6. Is the process stable and understood well enough to be accelerated?

7. What AI actions require human review, approval, or escalation?

   
   

These are not anti-AI questions. They are what responsible AI adoption looks like. Weak answers do not mean AI must be prohibited until every project practice is perfect. Perfection is not the standard, diligence and governance are.

They do mean AI should be introduced deliberately. Use it to help clarify the system, remove waste, strengthen traceability, and support better decisions.

AI is an amplifier, both good and bad

AI introduces genuine new risks involving access, scale, visibility, review, and automated action. The original RACI article is right that those risks require deliberate controls and named human accountability. But no AI governance framework can compensate for work that lacks a clear purpose. A RACI can identify the person accountable for an outcome. It cannot define what that outcome should be.

AI will magnify the strengths of a clear, disciplined system. It will also magnify ambiguity, weak governance, unnecessary work, and misalignment. Lean and project management lead to the same conclusion: establish purpose, value, success, traceability, and accountability; then understand and improve the work before asking technology to accelerate it.

The most important AI-readiness question may not be, “What can the tool do?” It may be, “What happens if it does exactly what we ask, faster and at scale?”

If the system is sound, that is an extraordinary opportunity.

If it isn’t, AI won’t create value; it will accelerate the waste already built into the system.

Before your next AI initiative, choose one process being considered for automation and examine its foundation. Ask whether the business need is understood, success has been defined, the work can be traced to value, accountability is clear, and sources of waste have been identified. If that examination exposes gaps, strengthen the process before asking AI to accelerate it. Automation should follow understanding, not attempt to replace it.

As organizations move more work to AI, which project-management fundamental do you believe deserves more attention? Share your perspective in the comments.

For organizations using SAFe, this sequence should be familiar: Lean-Agile thinking begins with value and flow, not automation for its own sake. If your organization wants to apply SAFe as a Lean-Agile system for strengthening project fundamentals and adopting AI responsibly, visit Prosper Project Management for more information.

#ArtificialIntelligence #AIGovernance #ResponsibleAI #ProjectManagement #ProjectGovernance #Lean #SystemsThinking #Leadership #ProcessImprovement #ValueDelivery #ProjectControls #ProsperProjectManagement