At least one IT firm believes enterprises will question scaling pilots as agentic systems demand more due diligence than bargained for.

1. Production shift
   After generative AI (GenAI) pilots proliferated in 2024–2025, enterprises this year will start to question large-scale deployment. Per an MIT Media Lab report, around 95% of GenAI investments yielded no returns. Going forward, production failures may expose gaps between prototypes and secure systems.

• Data access risks
  Following trends about data exposure concerns in autonomous AI, including prompt injection attacks documented in 2025, the firm warns against feeding models raw data, and advocates tool-directed filtering to limit exposure, costs, and errors.
  
• Context management
  AI in 2026 will struggle with stateless interactions and context switching compared to humans. Emerging practices will extend beyond prompts to organize metadata and history dynamically, as noted in 2026 trend analyses.
  
• Multi-agent orchestration
  Multi-agent systems will gain traction, although scale will remain limited. Protocols such as (model context protocol) will enable agent-to-agent communication for specialized workflows. Event-driven coordination is predicted to improve efficiency over single-agent limits.

Taking the AI road less travelled?
Funnekotter believes that the “flashy phase of AI is peaking: “In 2026, AI will start having to earn its keep. AI’s competitive edge isn’t just in better models or smarter prompts, it’s in connecting AI to the live, operational pulse of the business from day one.”

If he is right, then readers can expect a year in which more enterprises will get down to the fundamentals of strong implementations that ensure data security, improve the context of every model, and allow agents to work together in a robust enterprise infrastructure.