Appier, an AI-native Agentic AI-as-a-Service company, has released new research focused on enhancing the reliability of Agentic AI systems and Large Language Models. The study introduces a framework to evaluate and improve how LLMs handle decisions under varying risk conditions, addressing critical challenges in enterprise deployment.
Quick Intel
Enterprises are shifting from AI copilots to autonomous agents, but reliability remains a major hurdle to widespread adoption. A 2025 McKinsey survey indicates that 62% of organizations are experimenting with AI agents, yet inaccuracy continues to be the top cited risk. Appier's latest research directly confronts this by developing methodologies that make autonomous decisions more trustworthy and quantifiable.
The core innovation is the Risk-Aware Decision-Making framework, which moves beyond simple correctness evaluations. It incorporates structured risk parameters—rewards for accurate answers, penalties for errors, and costs associated with refusals—to mirror real-world enterprise trade-offs. Models assess their own capability and confidence before choosing to answer, refuse, or guess, with performance measured by maximization of expected reward.
Analysis reveals that many leading LLMs exhibit inconsistent behavior across risk levels. In high-stakes situations, they tend to guess excessively despite potential downsides, while in lower-risk contexts, they refuse answers too conservatively. This imbalance undermines both the autonomy and safety required for enterprise applications, highlighting issues in integrating knowledge, confidence, and risk assessment.
To overcome these limitations, the research proposes Skill Decomposition, a three-step reasoning process: first executing the task to produce an initial answer, then estimating confidence in that output, and finally reasoning about expected value under specific risk conditions. This method enables models to balance capabilities more effectively, resulting in rational, stable decisions particularly suited to high-risk enterprise scenarios.
"For Agentic AI to operate in critical enterprise workflows, the key is not only making AI smarter, but making its autonomous decisions more reliable," said Chih-Han Yu, CEO and Co-founder of Appier. "Appier has built its products around AI and continuously invested in world-class research. By turning LLM risk awareness into a quantifiable methodology, this research strengthens the foundation for trustworthy enterprise AI and helps accelerate the real-world adoption of Agentic AI and translate it into scalable business value and ROI."
Appier has applied these research insights directly into its Agentic AI-powered platforms, including Ad Cloud, Personalization Cloud, and Data Cloud. This integration supports enterprises in implementing more reliable autonomous marketing and data workflows. The company plans to continue advancing Agentic AI through its research strengths, proprietary data, and industry expertise.
About Appier
Appier (TSE: 4180) is an AI-native Agentic AI as a Service (AaaS) company that empowers business decision-making with cutting-edge AdTech and MarTech solutions. Founded in 2012 with the vision of "Making AI Easy by making software intelligent," Appier endeavors to help businesses turn AI into ROI with its Ad Cloud, Personalization Cloud, and Data Cloud solutions. Now Appier has 17 offices across APAC, the US and EMEA, and is listed on the Tokyo Stock Exchange. Visit www.appier.com for more company information, and visit ir.appier.com/en/ for more IR information.
TechIntelPro empowers B2B technology decision-makers with the latest industry trends, deep insights, leadership perspectives and actionable intelligence to drive both organizational and professional milestones—and thrive on the tech pulse.