BD
Buwaneka De SilvaAgile Delivery Leader
← Back to Insights
AIDeliveryResearch

Generative AI in Project Delivery: Promise vs. Reality

January 18, 2025·2 min read

Research Background

For my MSc dissertation at the University of Bedfordshire, I investigated how Generative AI tools (ChatGPT, Copilot, Gemini) are influencing Agile project delivery outcomes — specifically among IT professionals in Sri Lanka.

The findings challenged some assumptions I had going in.

What Project Managers Are Actually Using AI For

Based on interviews and surveys across 60+ IT professionals:

  • 40% use GenAI primarily for documentation generation (meeting notes, status reports, PRDs)
  • 28% use it for code review and technical specification drafting
  • 18% use it for risk identification prompting
  • 14% use it for stakeholder communication drafting

The pattern is clear: AI is being adopted as a cognitive offloading tool, not a decision-making partner.

Where It Helps (And Where It Doesn't)

GenAI genuinely accelerates delivery in documentation-heavy environments. A status report that took 2 hours now takes 20 minutes. Risk registers get first drafts in seconds.

Where it struggles:

  • Context retention — AI doesn't know your program history, team dynamics, or organizational politics
  • Nuanced judgment — Prioritization decisions require stakeholder alignment AI can't replicate
  • Trust gaps — Stakeholders want human accountability; they'll push back on AI-generated estimates

The Delivery Manager's New Role

The project managers who benefit most from GenAI aren't replacing their judgment — they're augmenting their time. By offloading synthesis and drafting, they free up cognitive bandwidth for relationship management, strategic alignment, and risk judgment.

That's exactly where human delivery leadership creates irreplaceable value.

The future isn't AI replacing project managers. It's project managers who use AI replacing those who don't.

← Back to all insights