Agentic AI: Navigating the Coming Cull to Identify Long-Term Winners

The AI revolution is hitting a crossroads. Over the next three years, Gartner predicts that 40% of agentic AI projects will be abandoned—a stark culling of hype-driven initiatives. Yet within this shakeout lies a golden opportunity: the rise of a select group of vendors and enterprises that will dominate the mature market. Investors must sift through the rubble of failed experiments to identify the 130+ firms with genuine agentic AI capabilities and the use cases that deliver tangible ROI. This is not just about surviving the purge but positioning for the 15% autonomous decision-making adoption rate by 2028, a milestone that will redefine enterprise software.

The Coming Cull: Why 40% of Projects Will Fail

The current AI landscape is a minefield of inflated expectations. Gartner's January 2025 survey revealed that 19% of organizations have already invested heavily in agentic AI, while 42% are proceeding cautiously—a sign of growing skepticism. The primary culprits for failure are clear:
- Escalating costs in training and maintaining large language models.
- Misaligned use cases, where AI is deployed for tasks better handled by simpler tools.
- “Agent washing”, where vendors rebrand RPA, chatbots, or basic automation as “agentic” without true autonomy.

The result? A wave of cancellations as companies realize their projects lack measurable returns. This purge will disproportionately impact the thousands of vendors that lack core agentic AI competencies, leaving only the 130+ firms Gartner identifies as having proven autonomous decision-making capabilities.

The Path to Long-Term Winners: Three Criteria for Survival

1. Avoid Agent Washing: Look for vendors that can demonstrate true autonomy—systems that make decisions without constant human oversight. Example: An AI agent managing supply chain disruptions by autonomously rerouting shipments and renegotiating contracts. Avoid firms peddling rebranded chatbots or RPA tools.

2. Target High-Value Use Cases: The most promising applications involve complex workflow automation—not incremental improvements. For instance, using agentic AI to redesign entire insurance claims processes, from fraud detection to payout approvals, rather than automating individual steps.

3. Legacy System Partnerships: Successful vendors will collaborate with enterprises willing to rebuild workflows from scratch to integrate AI. Legacy systems that bolt on AI to old processes will fail; the winners will be those that redesign systems to exploit AI's full potential.

The Long-Term Prize: 15% Adoption by 2028

The cull is temporary. By 2028, Gartner projects that 15% of daily work decisions will be made autonomously by AI agents, a leap from near-zero in 2024. This shift will drive demand for tools that:
- Augment enterprise productivity, not just individual tasks.
- Minimize hallucinations and errors via robust governance and synthetic data training.
- Upskill workforces—companies investing in AI literacy could see 20% higher revenues by 2026.

The prize is clear: firms that dominate agentic AI's infrastructure (e.g., cloud providers with scalable training frameworks) and those with use-case-specific solutions (e.g., autonomous legal research or medical diagnostics) will thrive.

Investment Strategy: Buy the Dip, Target the 130

The current shakeout creates a buying opportunity for selective investors. Focus on:
- Proven vendors: Companies like C3.ai (AI for industrial automation) or UiPath (advanced RPA with AI integration) that already show ROI in complex workflows.
- Enterprise software leaders: Firms like Microsoft (Azure AI) or Salesforce (Einstein) that are embedding agentic AI into legacy systems.
- Niche innovators: Smaller firms with vertical expertise (e.g., LatticeFlow for healthcare workflows or Akin for legal AI).

Risks and Governance

Investors must demand transparency: Ask vendors to detail specific use cases with quantifiable outcomes and risk management protocols. Avoid firms without clear paths to ROI. The 10% of boards relying on AI by 2029 will demand accountability, so governance and data quality are non-negotiable.

Conclusion: The Cull is a Filter, Not an End

The next 18 months will separate the signal from the noise in agentic AI. Investors who focus on autonomy over buzzwords, strategic use cases over incremental gains, and enterprise rebuilds over band-aids will be positioned to capitalize when adoption takes off. The winners will be the 130+ firms Gartner highlights—those with the grit to survive the cull and the vision to dominate the future of decision-making.

The time to act is now. Buy the dip, but buy wisely.