Home › News › AI Daily October 28 Afternoon Edition

AI Daily | October 28, 2026 Afternoon Edition: AgiBot GenieSim 2.0 Ignites Embodied Intelligence, Ant Group's Lingguang 2.0 Personal Agent Goes Live

Published: 2026-10-28 16:00  |  Category: AI Daily  |  Read: about 7 min

Top Stories This Afternoon

  • AgiBot launches GenieSim 2.0: the embodied-intelligence simulation platform reaches 96.7% real-robot consistency, with 200+ deformable body models and 50,000 datasets
  • Ant Group's Lingguang 2.0 personal agent goes live: stronger task hosting, as AI agents move from conversation to execution
  • NeuroHawk Technology raises RMB 700 million: the brain-computer interface track keeps heating up as invasive BCI enters clinical validation
  • Oracle diversifies its chip lineup: adding Google TPU and Cerebras to reduce sole dependence on NVIDIA
  • Microsoft's MAI model arrives: a new-generation multimodal AI model aimed at enterprise applications

1. AgiBot Launches GenieSim 2.0, Bringing a "Digital Twin Revolution" to Embodied Intelligence

AgiBotGenieSim 2.0Embodied Intelligence

AgiBot has officially released the GenieSim 2.0 embodied-intelligence simulation platform, describing it as not an ordinary iteration but a key step toward the "digital twin revolution" in embodied intelligence.

  • Supports 12 categories of non-rigid body modeling, with real-robot consistency reaching 96.7%
  • Ships with 200+ deformable body models and 50,000 training datasets
  • Engine code will be officially open-sourced on November 15, 2026
  • Can simulate complex industrial scenarios such as soft-body grasping and flexible assembly
Editor's note: Embodied intelligence is moving from the lab to industrial deployment. GenieSim 2.0's 96.7% real-robot consistency means simulation-trained results can be transferred almost directly to real robots, sharply lowering data-collection costs. The open-source plan is a shot in the arm for the entire ecosystem.

2. Ant Group's Lingguang 2.0 Personal Agent Goes Live as AI Agents Move from Conversation to Execution

Ant GroupLingguang 2.0AI Agent

Ant Group has officially released the Lingguang 2.0 personal agent product, substantially strengthening its task-hosting capabilities over version 1.0.

  • Supports cross-app task execution: automatically completing complex action chains such as ordering food, buying tickets and booking appointments
  • Adds a personal knowledge graph: builds personalized decision models based on user habits
  • Introduces a secure sandbox mechanism: operations involving payments or privacy require a second confirmation
  • Already connected to more than 500 lifestyle service providers across the Alipay ecosystem
Industry impact: The shift from "conversational AI" to "execution AI" is accelerating. Lingguang 2.0's direct connection to the payments ecosystem means an AI agent now has a complete commercial loop for the first time—it can not only help you decide, but also help you pay.

3. NeuroHawk Technology Raises RMB 700 Million as the Brain-Computer Interface Track Keeps Heating Up

NeuroHawk TechnologyBrain-Computer InterfaceFunding

China's leading brain-computer interface company NeuroHawk Technology has completed a new RMB 700 million funding round, with the proceeds mainly going toward clinical validation and commercialization of invasive brain-computer interfaces (BCI).

  • The company has successfully completed multiple invasive BCI implantation surgeries in humans
  • Its core product enables scenarios such as thought-controlled robotic arms and text input
  • The funding will accelerate clinical applications for conditions such as motor impairment and ALS
  • Brain-computer interfaces are seen as the next breakout point at the intersection of AI and neuroscience

4. Oracle Diversifies Its Chip Lineup with TPU and Cerebras to Cut NVIDIA Dependence

OracleChip StrategyTPUCerebras

Oracle is advancing a diversified AI chip strategy, planning to bring Google TPU and Cerebras wafer-scale chips into its cloud infrastructure to reduce its sole reliance on NVIDIA GPUs.

  • Has reached a TPU access agreement with Google Cloud and will offer TPU v5 instances on OCI
  • Is also evaluating Cerebras WSE-3 for large-scale model training
  • The move reflects how seriously large cloud vendors weigh the security of the AI compute supply chain
  • The global AI chip market is shifting from "NVIDIA dominance" toward "diverse coexistence"
Supply chain watch: From Oracle to Microsoft and Amazon, leading cloud vendors are actively looking for alternatives to NVIDIA. Not because NVIDIA isn't good enough, but because AI compute demand is growing far faster than any single supplier can expand capacity.

5. Microsoft's MAI Model Debuts, Targeting Enterprise Multimodal AI

MicrosoftMAIMultimodal

Microsoft has officially unveiled MAI (Microsoft AI), a new-generation multimodal AI model built for enterprise scenarios.

  • Supports unified understanding and generation across text, image, audio and video
  • Deeply optimized for Office 365, Azure and Dynamics 365
  • Introduces a zero-trust architecture for enterprise data security, so model inference never leaves the customer's tenant
  • Internal tests already cover automatic meeting minutes, contract review and financial report analysis

6. AI Accelerates Into Public Services as Beijing's Education Commission Issues 2026 AI Education Rules

AI RegulationEducationBeijing

As large models push into education, healthcare and other public-service scenarios at unprecedented speed, regulators are following close behind. The Beijing Municipal Education Commission has issued 2026 rules for AI use in education.

  • Defines the boundaries of AI use in homework grading and academic assessment
  • Requires schools to establish transparent AI usage mechanisms, with parents holding a right to know
  • Bans using AI assessment results as the sole basis for student progression
  • Encourages schools to offer AI literacy courses that teach students to use AI critically

7. Guizhou Releases a Centralized List of AI Application Scenario Needs

GuizhouAI ApplicationsLarge Models

Guizhou Province held an event to release results from AI industry large models and to match supply with demand, publishing a batch of AI application scenario need lists.

  • Promotes AI deployment under the slogan "looking for scenarios? Go to Guizhou"
  • Releases industry large-model applications such as Xiaotao Pharmacist and UniResearch
  • Covers distinctive local industries including healthcare, agriculture and tourism
  • Aims to build a demonstration hub for AI applications in western China

Today's Key Figures

MetricFigure
GenieSim 2.0 real-robot consistency96.7%
GenieSim 2.0 deformable body models200+
GenieSim 2.0 training datasets50,000
GenieSim 2.0 open-source dateNovember 15, 2026
NeuroHawk Technology funding this roundRMB 700 million
Lingguang 2.0 connected service providers500+
Ant Group Lingguang 2.0 supported modesCross-app task execution
Oracle's new chip suppliersGoogle TPU + Cerebras

Sources

  • Jiemian News: GenieSim goes viral! AgiBot's embodied-intelligence simulation platform upgraded, with 96.7% real-robot consistency setting the 2026 AI developer world ablaze
  • CSDN AI Industry Daily: October 28, 2026 AI industry daily | Genie 4 world model released, Oracle diversifies its chip lineup, Lingguang 2.0 personal agent goes live
  • Sina AI Hotspot Hourly: AI hotspot digest, October 28, 2026
  • China News Service Guizhou: Guizhou releases a centralized list of AI application scenario needs
  • Toutiao AI Daily News Brief: October 28, 2026, the AI industry breaks out on multiple fronts at once