Alibaba’s Wukong bets on agentic AI tokens

Large platforms are shifting from single chat assistants to networks of autonomous agents. This episode looks at Alibaba’s new Wukong system and the company’s Token Hub unit, a move that signals how AI may soon run quietly in the background of daily work while billing for each action it performs.

For operators, this change affects how teams plan, how work gets approved, and how software costs behave over time. It also raises questions about control and transparency as AI systems begin coordinating tasks across entire workflows.

We also cover Gecko Robotics’ major Navy contract, new MicroLED data center optics from MediaTek and Microsoft, and a new AI marketplace designed to help small and mid-sized businesses choose the right tools.

Learn more at crestvale.io

Transcript

SPEAKER_00 0:00

Welcome to CrestVail. This is a daily briefing breaking down what's happening across business, technology, automation, and why it matters. Today we're looking at how major platforms are shifting toward agentic AI that runs in the background of daily work. The shift is becoming harder to ignore. AI is moving from a single assistant you call on when needed to a quiet layer that runs tasks on its own. That change will reshape how teams work and how software gets priced. Markets closed higher in the previous session. The SP moved up. The NASDAQ also ended the day in positive territory. The 10-year yield drifted lower. Bitcoin continued climbing by the close. The center of today's story is Alibaba. The company introduced a new system called Wukong. It is built around a simple idea. Instead of one assistant doing everything, many small agents coordinate work across the tools people already use. Wukong edits documents without being asked. It approves routine requests. It summarizes meetings. It helps with research. It lives inside the places where teams already talk and plan. This is not meant to be another chat window. It is meant to be the layer between people and their tools. Alibaba is also building a new business unit called Token Hub. This unit will meter AI usage instead of charging flat fees. Agent systems tend to run long chains of tasks. Those chains consume more compute than a single request. Usage pricing helps Alibaba align cost with activity. Support for platforms like Slack, Microsoft Teams, WeChat, and DingTalk suggests Alibaba wants Wukong to be present wherever teams already work. If that happens, the AI becomes part of the workflow, rather than a separate tool. This matters because the model of how AI fits into work is changing. Operators will feel this shift in cost, in control, and in how teams design their processes. When AI runs on its own, companies need clearer rules around what it can do, who approves the work, and how much those actions cost. It turns AI from something optional into something more like utilities. Always on. Always billing. Always connected to the flow of work. In other news, Gecko Robotics secured a five-year deal with the Navy. The contract is valued at$71 million. Gecko will deploy climbing robots and AI software across 18 ships in the Pacific Fleet. These machines scan steel surfaces, map corrosion, and build detailed digital twins of each vessel. The Navy has been struggling with long maintenance cycles. These robots help teams inspect faster and with more precision. For operators and other industries, the lesson is direct. AI-driven inspections reduce downtime and reveal problems before they turn into major failures. MediaTech and Microsoft Research introduced a new optical cable built with micro LED technology. It replaces traditional laser-based components with hundreds of parallel light channels. The design cuts power use by as much as half and boosts reliability to levels similar to copper. Data centers running large AI clusters are hitting limits in power and heat. More efficient links give operators breathing room without redesigning entire network. The Proptery launched a marketplace for small and mid-sized businesses trying to adopt AI. It includes a free AI strategy director named Jordan. Jordan guides users through a structured conversation and returns a clear plan with recommended tools. Many teams feel overwhelmed by the explosion of AI software. This model shifts discovery from guesswork to a focused session that produces a next step. Vendors can list their tools. Enterprises can request deeper help. The goal is to make AI adoption feel doable rather than chaotic. Here's what else is worth knowing today. Sentinel 1 is seeing fast adoption of its purple AI analyst. Security teams are leaning more on autonomous tools to respond to threats as soon as they appear. Native raised$31 million to automate multi-cloud security policy work. As cloud environments keep expanding, many teams are facing more configuration drift than they can track manually. NIV AI came out of stealth with$12 million in funding. The company helps data centers recover wasted GPU power by predicting brief spikes in demand. This supports steadier performance in large training clusters. Pixart launched an AI agent marketplace aimed at creative teams. It offloads repetitive tasks like resizing assets and producing variations. This is another sign that agent workflows are spreading into everyday marketing work. Here's the operator takeaway. The moment AI moves from something you call on to something that runs in the background, cost, control, and workflow design all shift at the same time. If this was useful, follow Crestvale Newsroom so you don't miss tomorrow's briefing. Thanks for listening.