TechCapsules

Google is rolling out a new Google Home update that adds features to Gemini for Home and fixes Nest camera bugs. Gemini now asks clarifying follow‑up questions when multiple devices match a command, shows contextual images on Nest Hub displays, and delivers faster timer, alarm and sleep responses while allowing users to edit voice input before sending. The Google Home app update resolves live‑streaming and stability problems for first‑generation Nest Cam Indoor, Nest Cam Outdoor, wired Nest [...]

Q-CTRL’s Boulder Opal platform has achieved fully autonomous, cold‑start bring‑up and gate calibration on QuantWare’s D‑Line superconducting quantum processors. In a demonstration on a 21‑qubit QuantWare Contralto‑D21 device, the physics‑informed AI state machine tuned five‑qubit feedline blocks from an uncalibrated condition to gate fidelities suitable for operation in under three hours, without any specialist intervention. The system addresses a major scaling bottleneck: dozens of interdependent control parameters—such as TWPA biases, qubit drive frequencies and flux pulses—must be [...]

Google has begun rolling out a new Google Home update that addresses several long‑standing issues and adds new features across its smart‑home ecosystem. The update fixes the “Live View” problem that prevented some Nest Cam devices, including Nest Hello doorbells, from streaming video, and allows users to remove offline, unmigrated cameras from the Home app. On Nest Hub smart displays, visual answers now include relevant images when helpful, enhancing on‑screen information. Early‑access Gemini users receive faster timers, [...]

Microsoft and Novo Nordisk’s joint team rebuilt a clinical‑trial data analysis agent, moving it from Azure Container Apps to a hosted agent model in Microsoft Foundry. The new architecture isolates each user session in a VM‑sandbox, provides on‑demand compute, and leverages Foundry’s built‑in scaling, tracing, evaluation and governance tools. By offloading container scaling, session lifecycle and networking to the platform, the team reduced operational overhead while preserving security controls such as access policies and private telemetry. The agent [...]

Microsoft has launched a new browser extension for Chrome and Edge that lets users toggle Bing’s AI‑generated overviews and chat functions on or off. The add‑on places a switch in the browser toolbar, enabling a permanent “AI‑off” mode for all Bing searches, while also offering a manual “‑ai” operator for single queries. Microsoft’s blog notes the move responds to research indicating users want control over AI involvement in each session. The extension automatically sets Bing as the [...]

Token created an AI browser agent named Rehearsal that intentionally violated its own security guidelines—no distinct identity, unrestricted access, and no traceability—to expose weaknesses in agent protection. The agent operated within users’ authenticated sessions, performing tasks like opening pages and filling forms, making its actions indistinguishable from human clicks. By analyzing this failure, Token identified gaps in current practices and highlighted the need for separate agent lanes, identity tagging, and scoped permissions. The company outlined how emerging [...]

Google announced that Gemini Intelligence will soon provide proactive, personalized suggestions on Android Auto, but only for Pixel users. The feature will connect data across apps, enabling actions such as sending a meeting location with a single tap when a friend asks for directions. The rollout is limited to Pixel 10 and Pixel 11 models that already support Gemini Intelligence. Google first revealed the plan earlier this year, and recent updates have prepared the platform for the

Plurai, a startup founded by former Nexar engineers, announced a $13 million seed round to improve AI agent reliability by applying simulation techniques used in autonomous driving. The company builds digital twins of a firm’s internal systems, mapping all agent interactions to generate synthetic data that trains smaller, open‑weight models tailored to specific business rules. Early tests with Nvidia and Nebius show these models outperform mid‑tier frontier models at lower costs, addressing the high failure rates and expense [...]