Agentic AI digest (Sept 1–5 2025)

Business process services enter the agentic era (Sept 1)

NTT DATA published a blog titled “Agentic AI: Welcome to a new era in business process services” that explains how goal‑driven AI agents are transforming traditional business process outsourcing. According to Basu Bhaumik, instead of following rigid scripts and rules, agentic AI systems can make decisions, adapt on the fly and deliver outcomes without constant human supervision. This shift toward “services as software” (SaS) means organizations will consume modular, API‑driven agents to run tasks such as invoice reconciliation or claims assessment. NTT argues that agentic business process‑as‑a‑service (BPaaS) will reduce labour requirements while improving agility and speed. The article likens agentic BPaaS to moving from a taxi service that always needs a driver to a self‑driving car that knows where to go, avoids traffic and adjusts routes in real time.

Read the NTT DATA article

The growing challenge of auditing agentic AI (Sept 2)

Professional association ISACA released an analysis titled “The Growing Challenge of Auditing Agentic AI.” The post notes that as AI systems evolve from passive assistants to autonomous agents that act on behalf of users, traditional audit methods no longer suffice. Agentic AI often makes decisions based on complex models and opaque contextual reasoning, making it difficult to trace accountability when errors occur. ISACA urges organizations to develop new governance frameworks that balance innovation with transparency and risk management. Auditors should focus on understanding how agents are trained, what datasets they rely on and how outcomes are validated.

Read the ISACA article

Updates from the generative and agentic AI working group (Sept 5)

The COUNTER organization’s working group on generative and agentic AI released a short update summarizing its activities. The group is exploring how autonomous agents will impact usage metrics and digital content reporting. They are assessing whether AI‑driven interactions (e.g., agents downloading articles or accessing databases on behalf of users) require new counting models or metadata standards. In this first update, the group outlined its objectives and invited feedback from publishers and librarians.

Read the COUNTER update

Key takeaways

During the first week of September 2025, industry voices emphasized both the opportunities and risks of agentic AI. NTT DATA showcased how goal‑driven agents can run entire business processes, promising faster and more flexible service delivery. At the same time, ISACA highlighted the auditing challenges posed by autonomous AI systems and called for updated governance practices. Meanwhile, the COUNTER working group began examining how agentic AI might affect digital metrics and reporting, signalling that the ripple effects of agentic technology will extend far beyond the AI and tech sectors.