ServiceNow Drops Billions on AI—But Will Automation Actually Deliver?

ServiceNow announced its acquisition of Moveworks, an enterprise AI specialist, for two point eight five billion dollars. This acquisition is expected to be finalized in the second half of 2025 and aims to enhance ServiceNow’s artificial intelligence and automation capabilities by integrating Moveworks’ AI assistant and enterprise search technology. Currently, ServiceNow is already serving nearly one thousand AI customers and has surpassed two hundred million dollars in annual contract revenue for its Pro Plus AI offering. Moveworks has successfully deployed its AI assistant to almost five million employees across various organizations, with ninety percent of its customers using the offering company-wide.

OpenAI launched a new suite of tools and APIs to assist developers in creating AI-powered agents more efficiently. This includes the Responses API, which integrates features from its existing APIs, allowing developers to leverage web and file search capabilities, as well as a computer use tool for automating tasks. The open-source Agents SDK enables developers to build and manage agents using OpenAI models or even non-OpenAI models. The Responses API is now available to all developers, with usage billed at standard rates.

Google announced the release of Gemma 3, claiming it to be the most powerful artificial intelligence model that can operate on a single graphics processing unit. This model, which supports over 35 languages, is designed for developers to create AI applications across various devices, from smartphones to workstations. Gemma 3 boasts enhanced performance, outperforming competitors like Facebook’s Llama and DeepSeek on single GPU setups. The model includes an upgraded vision encoder that can handle high-resolution and non-square images, along with a new safety classifier to filter harmful content. Despite its advanced capabilities, Google emphasizes that Gemma 3 has been evaluated for potential misuse, indicating a low risk level for creating harmful substances. The company is also offering Google Cloud credits to promote Gemma 3, while academic researchers can apply for a program providing them with ten thousand dollars in credits to support their research endeavors.

Alibaba Group introduced a new artificial intelligence model capable of reading emotions, aiming to compete with OpenAI’s latest offerings. The R1-Omni model, developed by Alibaba’s Tongyi Lab, was demonstrated to infer emotional states from video, while also providing descriptions of a person’s clothing and surroundings. This model enhances computer vision capabilities and is an improved version of the previous HumanOmni model, created by lead researcher Jiaxing Zhao.

Why do we care?

Two threads.  First, AI-driven automation is the current bet.    Second, models are getting smaller and having specialized focus.   A single-GPU AI model that outperforms competitors like Llama is a big deal. It suggests that AI workloads are becoming more accessible, allowing more businesses to run powerful AI on lower-cost infrastructure. 

While these advancements are impressive, they also highlight the risks of AI hype. AI-powered automation tools promise efficiency, but their implementation often falls short of expectations.  Filter based on aligning AI adoption with real business outcomes rather than chasing trends.