📍UBS Counts 12,700 Personal Agents
Plus: Help Before the Appointment, Home Depot Checks the AI Findings
Good Morning, Agent Owners!
The first AI-run company will have no employees, no office, and somehow still require mandatory team-building activities.
UBS Counts 12,700 Personal Agents
TL;DR: UBS says employees are running more than 12,700 personal AI agents to support daily work. Its third-quarter results separately report 608 AI use cases in production and more than 18,000 developers using AI tools. The disclosure shows how broadly agents have entered the bank, without establishing how many people use each agent or isolating their contribution to financial performance. Read more →
Which Meeting Spiked Your Stress?
TL;DR: Shopify CEO Tobi Lütke connected his wearable data to a personal agent and asked it to identify his most stressful meetings, interviewer Alex Heath reports. He then told the people involved, who apparently did not appreciate the finding. It is a glimpse of agents connecting health and calendar data, though the anecdote does not establish what caused the stress. Read more →
Help Before the Appointment
TL;DR: Relationships Australia SA uses agents to screen clients while they wait five to six weeks for appointments. An agent reviews risk questionnaires and flags concerns for trained staff, who can call within hours or days. Another prepares summaries for practitioners. The nonprofit describes an active workflow, but has not published error rates or evidence that the system reduces harm. Read more →
Utah Puts Coding Agents to Work
TL;DR: Utah gave coding agents to 34 developers across 10 state agencies in a four-week pilot. Anthropic says 71% reported saving more than 20 hours, while 80% had a working environment within 30 minutes. The results suggest a practical starting point for public-sector adoption, although they are supplier-reported participant estimates rather than independently measured productivity gains. Read more →
Home Depot Checks the AI Findings
TL;DR: Home Depot’s CIO says AI security models found issues in the retailer’s code that needed fixing, alongside false positives. The company has joined an advanced AI security program and is reconsidering how quickly it patches vulnerabilities. Her target is to bring remediation down to hours. The account shows real adoption, while leaving people responsible for evaluating what the models report. Read more →
Mercor Leaves the Long Tasks Running
TL;DR: Mercor engineers assign long development tasks to agents before bed, during meetings or on the way to the gym, according to a supplier’s customer account. They return to tested features and pull requests ready for review and merging. The workflow shifts execution into otherwise unavailable hours, but the account provides no independent productivity measurement or basis for attributing company growth to agents. Read more →
The First Draft Goes to Agents
TL;DR: GitHub says more than 40% of pull requests in September began with an agent’s draft, up from 17% in March. That amounted to roughly 110 million agent-started pull requests in one month. The platform’s annual report illustrates a shift toward directing and reviewing agents. The figures measure drafting involvement, rather than autonomous delivery or independently verified code quality. Read more →
Congress Questions the AI-Run Company
TL;DR: The U.S. House Oversight Committee is seeking documents about Delaware’s proposed companies managed by AI agents. Its chair questions potential liability protections, safety assessments and private-sector influence over the plan. The October 26 inquiry adds federal scrutiny to an existing proposal. It does not mean Delaware has enacted the framework or that agent-run companies have received blanket legal immunity. Read more →
Balance Checks Take Seven Percent
TL;DR: Southern Cross Credit Union has deployed a phone agent that verifies callers, connects to its core banking system and provides real-time account balances. The agent now handles 7% of all calls, according to its supplier’s case study. Built in six weeks, it targets a narrow, frequent request. The account does not quantify the resulting reduction in customer waiting times. Read more →
More Security AI, Few Job Cuts
TL;DR: In an a16z survey of 100 CISOs at companies with more than 1,000 employees, nearly half expect security headcount to rise in 2027 and only 4% expect cuts. Internal agent security and security operations automation feature prominently in buying plans. The findings suggest expanding coverage can absorb automation gains, but these are executives’ intentions rather than observed employment outcomes. 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👇
