LLMs

SS&C Blue Prism is guiding clients from RPA to agentic AI, a necessary evolution for complex workflows. Traditional RPA struggled with unstructured data, while agentic AI, leveraging LLMs, can reason and adapt in real-time. SS&C Blue Prism focuses on an outcome-oriented approach, setting goals rather than dictating steps. While fully autonomous AI is still developing due to trust and regulatory concerns, SS&C Blue Prism is introducing new technology to embed AI agents into existing workflows, aiming to unlock significant further automation potential.

Separating core agent logic from execution strategies is crucial for scalability. Researchers propose Probabilistic Angelic Nondeterminism (PAN) and the ENCOMPASS framework, which allows developers to define the “happy path” of an agent’s workflow while deferring inference-time strategies to a runtime engine. This decoupling reduces technical debt and enhances performance, enabling independent optimization of logic and search algorithms without code modification.

MiniMax Group Inc. debuted on the Hong Kong Stock Exchange with a 90% surge, raising HK$4.8 billion. This AI startup, specializing in large language models (LLMs), is the second major Chinese LLM developer to go public, following Zhipu AI. Backed by Alibaba and Tencent, MiniMax aims to compete with U.S. AI giants. The IPO funds will boost R&D, underscoring investor confidence in China’s AI sector despite geopolitical hurdles and semiconductor import restrictions.

Boards are demanding productivity gains from enterprise AI, but features like web browsing and application connectivity introduce cybersecurity risks, including indirect prompt injection attacks. Tenable research highlights these vulnerabilities, potentially enabling data exfiltration and malware persistence. Mitigation requires treating AI assistants as distinct IT entities, subject to rigorous audit and zero-trust controls, including a comprehensive AI system registry and context-aware feature constraints. Organizations must invest in training and continuous monitoring to proactively address emerging threats and evolving vendor security postures.

ServiceNow CEO Bill McDermott addresses concerns about AI potentially displacing enterprise software. He emphasizes ServiceNow’s integration with major AI hyperscalers, viewing it as a collaborative ecosystem rather than a threat. McDermott argues that AI models won’t replicate ServiceNow’s comprehensive solutions for business processes, especially in complex environments. He highlights the limitations of siloed AI deployments and stresses the importance of a cross-functional approach. Additionally, ServiceNow announced a five-for-one stock split to attract retail investors, following strong earnings that exceeded expectations.

In a benchmark test simulating social dynamics, seven LLMs played the game Werewolf. GPT-5 significantly outperformed the others with a 96.7% win rate, demonstrating superior strategic thinking and manipulation skills. Other models, including Qwen3 and Kimi-K2, showed respectable performance. Analysis revealed distinct personality traits in each model; for example, Kimi-K2 exhibited aggressive tactics. The experiment highlights the importance of social skills for AI agents operating within human teams, alongside traditional benchmarks.

At the AGI Playground, Liang Zhihui of 360 Group introduced “Super Search,” an AI-driven search engine. Moving beyond traditional keyword-based methods, Super Search uses a task engine with autonomous planning to address complex queries. It leverages LLMs, along with specialized tools and models, including domestic ones, to break down and execute intricate tasks, as for planning a rock-climbing trip. Features include intelligent agent creation and integration of high-quality models.