By Raymundo Tamayo, a Humentum Associate
Why it matters now
As we move into 2026, artificial intelligence is no longer a novelty in project work, it’s widely adopted but unevenly impactful. According to McKinsey’s State of AI: Global Survey 2025, nearly nine out of ten organizations now use AI in at least one business function, and many are experimenting with or beginning to scale AI agents (autonomous systems that can plan and act across workflows). Yet only a minority have moved beyond pilots into durable delivery value.
This presents a clear challenge and opportunity for project managers: AI tools can accelerate workflow, but real project value comes from integrating AI into delivery systems, aligning it with outcomes, risk management, and responsible adoption.
Why 2025 was a turning point
In early adoption cycles, attention is focused on individual tools and simple prompts. Teams scramble to try the latest generative AI models, to draft documents, summarize meetings, or prototype ideas. But the evidence shows this alone doesn’t move performance metrics. The basic questions are no longer “Can we use AI?” but rather “How do we embed AI in how we deliver work and manage change?”
- McKinsey finds widespread use but persistent limitations in scaling, most organizations remain in experimentation or pilot phases, not enterprise-wide delivery.
- Independent research indicates most generative AI pilots fail to deliver measurable financial or productivity outcomes, with only a small minority yielding sustained impact.
For teams in social impact organizations, nonprofit networks, and public sector delivery partners; this pattern is familiar: promising ideas often stall when they encounter complex data environments, stakeholder expectations, compliance requirements, and trust issues. Project managers are uniquely positioned to help teams move from experimentation to routine delivery.
The Five Foundations for Using AI Well
To bridge the gap between tools and value, project managers should focus on five interrelated areas. Each is necessary, but none is sufficient on its own.
1. Discern what tools can and cannot do
AI isn’t a monolith. A chatbot is not an analytical agent; a draft generator isn’t a decision engine. Spend time matching tools to task needs and identify their limitations clearly before integrating them into delivery steps.
2. Shift from prompts to workflows
Individual prompts help with convenience; integrated workflows drive outcomes. Project teams benefit when AI supports repeatable processes: planning cycles, status updates, risk reviews, or compliance checks.
3. Design pilots to scale (or stop fast)
Generic pilots without success criteria, adoption plans, or owners rarely deliver value. Treat AI pilots like any other project initiative: define clear objectives, metrics, and a transition path to production or termination.
4. Build the skills that matter most
AI won’t replace project managers, but it will reshape the capabilities that drive impact. In development and ESG work, priority skills now include adaptive program leadership grounded in real-time evidence, responsible data stewardship, impact-focused measurement, ethical AI governance, and inclusive stakeholder engagement. These competencies help teams use AI safely, protect communities, and strengthen delivery outcomes.
5. Establish Governance that enables speed and trust
Shadow AI (employees using unvetted tools) is a growing risk. Rather than banning AI, teams benefit from lightweight guardrails that clarify data use, documentation, output validation, and escalation paths so innovation doesn’t outpace safety.
A simple way to start today
Before you dive into a specific tool or platform:
- Inventory current AI use cases in your projects (what’s used, why, who owns it).
- Map where these sit in your delivery workflows (planning, execution, reporting).
- Ask one strategic question: “Does this reduce friction in delivery and align with risk expectations?”
Answering this starts the shift from experimentation to integration, the practical heart of high-value AI use.
How this series will help you
Over the next weeks, we will publish a five-post series that dives into each of these moves with practical frameworks, simple templates, and clear examples project managers can adopt immediately. This series is designed to be used as a practical reference, not just read once. Each article addresses a specific decision area in AI-enabled project delivery.
Here are the topics we will cover:
- Know What AI Tools Can and Can’t Do
- From Prompts to Workflows
- Escaping Pilot Purgatory
- Five AI Skills Project Managers Should Build
- Governance That Enables Trust and Speed
References
- McKinsey – State of AI 2025 (Nov 2025)
- MIT NANDA – “The GenAI Divide: State of AI in Business 2025” (Aug 2025)
- ISACA – “The Rise of Shadow AI: Auditing Unauthorized AI Tools in the Enterprise” (Sep 2025)
- BCG – “How Agentic AI Is Transforming Enterprise Platforms” (Oct 2025)
- Atlassian – Nonprofit AI Collaboration Index 2025 (Sep–Nov 2025)