📌 Plaud launches AI pin and desktop meeting notetaker

The $179 NotePin S comes with a clip, lanyard, magnetic pin, and wristband, supports Apple Find My, has 64GB storage, 20-hour battery, and dual MEMS mics capturing clear audio within 9.8 feet.

Plaud’s desktop app detects active meetings, records system audio, and uses AI to structure transcripts into notes, also supporting multimodal inputs like images and typed annotations.

With over 1.5M devices sold, Plaud targets both mobile users and virtual meetings, positioning itself against transcription apps like Granola, Fathom, and Fireflies.

Counterforce Health built an AI platform to help patients and clinics challenge denied coverage requests.

Manual appeals were slow, requiring hours of interpreting dense policies, reviewing codes, and drafting letters, with most requests going unsubmitted.

The AI analyzes denial letters, patient records, and policy details, then drafts targeted appeal letters that meet provider requirements.

Clinics using the system saw preparation time drop from hours to minutes, with about 70% of appeals approved, reducing administrative strain and speeding reimbursements.

Alaska’s court system spent over a year developing an AI chatbot, AVA, to guide residents through probate filings. Originally planned as a three-month project, it repeatedly produced inaccurate or incomplete answers.

Early versions of AVA struggled with hallucinations, overly empathetic or confusing tone, and referencing incorrect sources. The team had to run extensive testing to ensure responses were accurate enough for residents navigating sensitive personal matters.

The AVA project demonstrates the gap between AI optimism and practical implementation in government services. Even with careful development, AI systems require continuous evaluation to maintain trustworthiness for public-facing applications.

A DoorDash customer said a driver marked an order as delivered using an AI-generated photo of his front door and a food bag that did not arrive. The image closely matched his real doorway, prompting questions about how the driver could replicate it without showing up.

The post triggered debate over DoorDash’s photo rules. Some drivers said uploads from the camera roll are blocked, while others claimed drivers can reuse past delivery images, making this scenario plausible.

It was later suggested that the delivery may have involved GPS spoofing, a modified cellular device, or reused images. AI tools are making it easier to bypass basic proof-of-delivery checks, pushing platforms to strengthen verification systems.

Samsung plans to double Gemini-powered devices to 800M this year

LG says its CLOiD home robot will fold laundry and make breakfast

Boston Dynamics’ AI humanoid is learning factory work