AI agents
Shopify opens its catalog to more AI shopping agents
Shopify says more AI agents can access over a billion product listings through Shop. Buyers can set shopping rules and must approve each purchase.
By The AI Feed Desk
Liquid AI releases LFM2.5-24B-A2B for fast on-device agents
Liquid AI's 24B mixture-of-experts model uses 2B active parameters for local reasoning and tool calls. Its agent workflows differ from d1's single-pass decisions.
By The AI Feed Desk
Anthropic rebuilt source control for 30,000 daily code merges
Anthropic replaced Git with Scion as Claude-driven code volume grew. Its case study offers lessons in checkout costs, server APIs and migration tradeoffs.
By The AI Feed Desk
Amex introduces a framework to stop AI agents making unauthorized purchases
Amex's Intent Intelligence Framework records approved purchase limits before an AI agent pays. Its protection depends on what the customer signs off on.
By The AI Feed Desk
Meta releases Muse Wonder, its first frontier model
Muse Wonder brings Meta's frontier-model push to the API. Wonder Full stays in Muse Code, so builders face different access and evaluation choices.
By The AI Feed Desk
Ex-Plaud executive raises $26M for Ambi, an AI app that listens in the background
Ambi launches iPhone, Mac and Apple Watch apps for ambient capture. Its promise of help during meetings raises a test of whether all-day listening adds value.
By The AI Feed Desk
TypeSafe, maker of non-text model Jev, reaches a $7.5B valuation weeks after launch
TypeSafe raises $870 million for Jev. Its enterprise adoption and savings claims leave buyers with a task-level test of cost, speed and decision errors.
By The AI Feed Desk
Satya Nadella says AI models should be assumed compromised and need an emergency brake
Microsoft's CEO calls for controls outside AI models, records of their actions and a human stop mechanism. Buyers can turn that proposal into a practical test.
By The AI Feed Desk
Anthropic cuts internal evaluations off from the internet after its AI sent a false homicide tip to Philadelphia police
Anthropic found a false police tip more than two months after submission. Its offline testing response exposes separate gaps in prevention and detection.
By The AI Feed Desk
Natura's $99 Interface smart ring summons AI agents with a press
Natura plans a ring for voice requests, device control and health tracking. Its $9 monthly subscription raises the first-year cost to $153 to $180.
By The AI Feed Desk
Goodfire launches monitors that look inside models to catch rogue AI agents more cheaply
Goodfire screens internal model signals before calling an AI judge. Its cost and detection results use different thresholds, a key detail for buyers.
By The AI Feed Desk
Cal AI's 19-year-old founder raises $10M for a personal AI agent startup
Zach Yadegari's Persona pairs an iMessage agent with a planned $179 band. Instinct's hands-on results show what to check before delegating purchases.
By The AI Feed Desk
Google turns Gemini into a business agent that delegates to subagents and has its own email address
Google's Gemini agent plans work across business apps and delegates to subagents. Its coworker accounts make permissions and audit trails a buying question.
By The AI Feed Desk
Nous Research confirms a $1.5B valuation and launches business agents
Nous Research raised a $90 million Series B for its business-agent push. Hermes adoption figures leave paying demand and workflow reliability to be tested.
By The AI Feed Desk
Microsoft and Nvidia launch RTX Spark PCs, with laptops priced up to $7,000
RTX Spark laptops start at $2,599. The $7,000 model is an Asus, while the buying case against a MacBook Pro depends on memory and the software you run.
By The AI Feed Desk
Meta's Muse agent comes to iPad a month after its mobile debut
Muse reaches iPad 29 days after its mobile launch. More connectors expand its work and shopping options, but retailers can still block the agent.
By The AI Feed Desk
Anthropic launches Claude Haiku 5.5, its fastest and cheapest small model
Haiku 5.5 cuts token prices for high-volume work. The 100,000-token threshold and Sonnet's cheaper cache reads change the model-choice calculation.
By The AI Feed Desk