Why Most AI Projects Stumble
Statistics from Gartner indicate that roughly 85% of AI projects fail. According to Welsch, this high failure rate often stems from a fundamental tactical error: starting with the technology rather than the problem. Leaders often feel pressured to “do AI” without aligning it to their established business strategy or key performance indicators (KPIs). For AI to be successful, it must speak the language of the business function it serves, moving beyond the quantity of agents deployed to focus on measurable impact—whether that is increased revenue or faster problem resolution.
The Evolution of the “Talent Bench”
One of the most pressing concerns in the current AI era is how it impacts organizational structure. Welsch notes a recent shift in corporate strategy, highlighting IBM as a key example. While IBM initially announced a reduction in back-office staff in 2023, the company later revealed a tripling of entry-level hiring. This highlights a critical realization: organizations must still build a “talent bench” of people who understand their customers and technology from the ground up to eventually fill senior roles.
Welsch explores these changing structures in his bestseller, The Human Agentic AI Edge, discussing how organizations are transitioning from traditional pyramids to “diamond” or “spear” shapes where a few leaders manage a vast network of AI agents.
Cultivating Subject Matter Expertise
Despite the automation of manual tasks, Welsch argues that subject matter expertise is more critical than ever. Without it, workers cannot effectively:
- Define Goals: Write high-quality instructions or prompts that align an agent with its intended purpose.
- Audit Output: Judge whether the results meet the organization’s bar for quality and accuracy.
- Think Critically: Develop the judgment necessary to know if information is actionable and correct.
To foster this expertise, Welsch recommends corporate “cross-pollination” through fellowships, job shadowing, and apprenticeships, allowing workers to understand how a company truly functions across different departments.
The “Magic Moment” of Personal Automation
For individuals looking to start their own AI journey, Welsch encourages hands-on experience. He shares his personal “magic moment”—automating his content repurposing workflow using tools like n8n and Stability AI. What previously took him six to seven hours of manual labor now requires only a few minutes of review, allowing him to maintain a high-quality podcast and newsletter while focusing on strategic tasks.
Final Word: Stay Present
