Startup Atmanity released Atmee v2, a real-time avatar model that generates lifelike video-callable digital humans from just one photo and 10 seconds of audio. Available via API, the tool aims to provide visual interfaces for AI agents in education and support, marking a shift from text/voice to video interaction.
Multimodal Interaction
多模态交互
Human-computer interaction combining visual, voice, and other sensory channels.
Curated Intelligence
Google is expanding its AI Mode advertising business to international markets to offset slowing growth in core search ads. While Q3 revenue rose 25% year-over-year, stock prices fell due to ad growth concerns. Early tests show sponsored lists in AI Mode match organic click-through rates, with AI queries averaging three times the length of traditional searches.
A team from the University of Tokyo introduced the MIDA benchmark for multimodal interactive deception assessment, accepted at EMNLP 2026. Tests reveal that leading multimodal LLMs like GPT and Claude struggle to reliably distinguish truth from lies in social interactions, despite processing audio-visual-text inputs. The study highlights significant gaps in models' ability to 'read the room' and understand theory of mind.
Rokid and Tencent Mobility launched the industry's first native AI glasses ride-hailing service. By connecting their agents via the A2A protocol, users can complete the entire booking-to-payment process using only voice commands on the glasses, significantly lowering interaction barriers.
Square has launched the Square Kiosk, a self-service ordering terminal featuring an agentic conversational voice layer. The device enables real-time menu understanding and natural language interaction for customers, aiming to replicate human staff service. It is priced at $1,999 plus a $99 monthly software fee.
XPeng has activated the production line for its IRON humanoid robots, targeting mass production by late 2026 and deliveries in H1 2027. The robot features multimodal interaction with trilingual support and context memory, while XPeng's second-generation VLA model bridges L2 to L4 autonomous driving capabilities, now deployed in their Robotaxi service.
Apple has scheduled a 'Welcome Home' event for October 13, expected to unveil its first smart home hub featuring a 6-inch display. The device is rumored to use an A18 or higher chip to support Apple Intelligence, incorporating human detection, gesture recognition, and FaceTime-based identity verification as the central node for home automation.
The Verge conducted a one-year deep-dive review of Amazon's Alexa Plus. The conclusion indicates that while the AI assistant performs reliably in smart home device integration (e.g., lights, coffee makers), it still suffers from significant shortcomings in long-term memory, personalized context understanding, and complex life task handling, with intrusive ads severely degrading user experience.
OpenAI has introduced an 'Intelligent UI' feature for ChatGPT, enabling the model to dynamically generate images, charts, and clickable buttons based on context. This update aims to transform text-only conversations into interactive visual experiences, enhancing information presentation efficiency.
openJiuwen has launched WorkSwarm, enabling full-duplex multimodal agent interactions similar to a phone call. This solution allows users to interrupt or share visuals while the AI executes complex background tasks, addressing latency issues in traditional turn-based dialogue. The core innovation lies in the dual-rhythm coordination between real-time models and the Core Agent, with native affinity for Ascend computing power.