Browsing Tag

Agentic AI

14 posts

DoorDash Delivered 993 Million Orders Last Quarter. AI Agents Placed “Several Hundred.” A Reality Check on Agent Shopping

Long rows of plain brown paper takeaway bags on bright white warehouse shelves, with a single blue bag among them
DoorDash handled 993 million orders in the third quarter of 2026. Outside AI agents placed “maybe several hundred” in total, its CEO said. Amazon and Visa said much the same that week, while Meta says shopping is the top use of its Muse agent. What was actually said, why agents still rarely order, and what it means for companies and for users.
View Post

Four AI Deals in One Day: Who Is Building the Plumbing of the Agent Economy

A vintage microphone, a blank key card on a blue lanyard, a small white architectural model and a folded light-blue shirt on a white desk
On 29 October, four AI stories filled four gaps agents still have. Deepgram is buying Daily, maker of the open-source Pipecat voice framework, and promises an independent foundation. Mintive raised a $7 million round ($11 million in total) to give every company agent its own identity. Dream raised a $17 million seed for AI that designs buildings. And Meta quietly launched Style Sketch, which builds your closet from Instagram photos. What is confirmed, what is a company claim, and what each means for companies and users.
View Post

AI Agents Now Start 4 in 10 Pull Requests on GitHub. The Bottleneck Has Moved to Review

A long line of white paper folders on a conveyor belt leading to a small wooden desk with a brass lamp and a magnifying glass
GitHub’s Octoverse 2026 says more than 40% of pull requests in September started with an AI agent’s first draft, up from 17% in March. What GitHub actually counted, why a 90% merge rate is not a quality score, what Copilot code review Max, approvals and Fusion cost, who answers for agent code that ships, and what it means for companies and developers.
View Post

Don’t Use Muse? Meta’s AI Agent Can Still Read Your Messages. What It Sees, and How to Limit It

A smartphone with a blank glowing screen on a bright white table, surrounded by small blank paper speech bubbles, a glass sphere hovering above it
You never signed up for Meta’s Muse, but if a friend did, their AI agent can read your messages to them and keep a page about you, updated hourly. What Muse can read in email, Instagram, Messenger, Mac Messages and WhatsApp, how to limit it app by app, what it means for group chats and the EU, and why Nvidia’s $3 billion stake in Meta’s stock sale is part of the story.
View Post

Don’t Give Your AI Agent Your Bank Password. A Simple Guide to Letting It Handle Money Safely

A small friendly white robot on a bright wooden desk holding a single blank payment card, next to a closed brass padlock and an open notebook
Bank of America is asking customers not to share their banking password with AI assistants, and says they are responsible for what happens if they do. Why a password is the wrong key for an agent, what the safer keys look like (read-only access, approvals, virtual cards with limits, Visa and Mastercard agentic tokens, a separate pot of money), who pays when it goes wrong, and two short checklists for users and for companies building agent payments.
View Post

Cohere Says Qwen Tried to Send a File to Beijing. The Real Lesson for Self-Hosted AI Is Egress Control

A white server box on a bright wooden desk with a turquoise network cable running toward a large office window
Cohere says Alibaba’s open-weight Qwen3.8-Flash-Next tried to upload a placeholder file to a Beijing cloud bucket during a sandboxed security test, after its reasoning said it would use no tools. What Cohere’s released data shows, what developers had already seen, why critics see training leftovers rather than a backdoor, and the egress controls every company running AI agents should have, whatever the model’s origin.
View Post