📍Publishers Prepare a Feed for Agents
Plus: The Phones Stay Open After Five, The Old Dashboard Stays Open
Good Morning, Agent Owners!
AI agents now get clean, distraction-free news feeds. Humans can continue enjoying cookie banners, autoplay videos, and seventeen subscription pop-ups.
Publishers Prepare a Feed for Agents
TL;DR: People Inc. is building custom APIs that give agents article text, authorship and publication or update dates without the surrounding webpage. Its innovation chief says sending unnecessary page elements wastes processing and distracts agents. The project shows a publisher adapting distribution for machine readers, with greater control over delivery. The company has not disclosed resulting revenue or measured cost savings. Read more →
The Phones Stay Open After Five
TL;DR: White River Hardwoods built six agents to handle calls, check live inventory, identify products and open trade accounts after its sales office closes. Conversations reach the right salesperson for follow-up. During business hours, agents step in after callers wait more than 60 seconds. The manufacturer's account describes an operating workflow, though it provides no measured increase in orders or revenue. Read more →
Cisco Funds a Legacy-Code Challenge
TL;DR: Cisco president Jeetu Patel gave engineer Padmanabhan Krishnamurthy unrestricted AI usage after he proposed tackling legacy code with a two-person team. Fortune reports an initial four-week effort to prove the approach, with a goal of refactoring 98% of legacy code within 24 months. The assignment shows executive backing for a small-team experiment; completion results and total costs remain undisclosed. Read more →
ADNOC Plans Connected Oil Decisions
TL;DR: ADNOC has signed a partnership to connect oilfield planning with production, refining, logistics, sales and trading. Interconnected agents would weigh operational and commercial decisions together, starting with field development plans and value-chain analysis. Human experts retain oversight of consequential decisions. The agreement sets a development agenda: preparing plans in hours instead of months remains a stated ambition, without published validation. Read more →
Food Imports Meet Machine-Readable Rules
TL;DR: Abu Dhabi says a production pilot is helping officials assess food imports against a defined set of machine-readable government rules. Each assessment can be traced to the relevant policy, with people retaining responsibility for final decisions. The deployment gives regulatory AI a concrete workflow, while leaving processing volumes, accuracy and time savings undisclosed. Wider government adoption remains an ambition. Read more →
A Surprise Gift Arrives
TL;DR: Justine Moore says she authorized an agent to choose and buy her a surprise gift, using a spending limit and access to her posts and prior conversations. A week later, she shared a photograph of the delivery. The account links personal context to a completed purchase, although it remains one user's experience rather than evidence of consistent shopping performance. Read more →
Claim Forms Become Reviewable Drafts
TL;DR: Definity has put agents to work on brokers' first notices of loss. The system classifies submissions, extracts information and creates structured claim drafts inside its claims platform, where adjusters review them. The supplier's customer announcement describes a production deployment that replaces manual retyping. It supplies no quantified efficiency result, and people remain responsible for the subsequent claims decisions. Read more →
The Old Dashboard Stays Open
TL;DR: A case study of 150 sales executives found that 82% of sales-intelligence queries moved to an agent during its first month. Yet 55% of users kept their dashboards for complex charts and reporting. The agent's adoption changed daily work without replacing every tool. The company is unnamed, and the author's reported results do not establish improved sales or causation. Read more →
150 Checks, One Review Workbook
TL;DR: Smarsh used an agent to inspect a real workload's code, architecture documents and tests against 150 best practices. In hours, it produced a workbook linking assessments to evidence, gaps and suggested fixes. Engineers then challenged findings and prioritized remediation. The company describes its first internal uses, including a second repository; broader rollout and continuous checks are still planned. Read more →
A Notification Tool Reaches 1,300 Inboxes
TL;DR: Mandiant describes an intrusion in which an agent abused an application's existing notification feature to send test emails to more than 1,300 internal users. It also changed records and attempted other actions through legitimate functions. The case shows how compromised access can turn ordinary business capabilities into attack tools. The victim and exact incident date have not been disclosed. Read more →
Your agent cloud, everywhere — in your chat apps, browser, and tools. Always on.
- Add to Slack, Teams, Telegram or WhatsApp.
- With a clean, redesigned WebUI.
- Connect any of 100+ apps & sources in one click.
- GPT 6 Astra, DeepSeek v4.1 & more frontier AI models, built in at discounted rates.
Run your agent on MyClaw.ai in Seconds👇
