Skip to content
UPDATED EACH WEEKDAY · 56 TAKES SINCE OCT 15 · LAST THU NOV 19

The H.O.T. Take

What changed in AI, read where people, organization and technology meet, and what it means for one workflow on Monday.

This week's lead · Moonshots robotics deep-dive: when machines outnumber staff, name who hands over each task · plays from its own channel

Today

Thursday · Nov 19 · 2 takes

Staffbase: frontline teams lose time to missing information, and want AI answers

A Forrester study for Staffbase finds 48% of frontline workers and managers say information gaps cut their productivity, only 23% feel fully supported, and 47% rank AI-powered guidance among the tools they want most. On Monday, take one frontline workflow, settle which source holds the right answer, and make sure the AI assistant and the manager point to the same one.

Microsoft's Agent 365 lists each agent's owner and flags the ones with none

Microsoft's Ignite update pulls agents from connected platforms into one registry with their owner and tools, flags ownerless or inactive agents for review, and can require a person's approval for sensitive actions as an agent's risk rises. On Monday, list the agents touching one workflow, give each a named owner, and decide which actions still wait for a human yes.

Earlier this week

11 takes

WEDNESDAY · NOV 18

Organization leads · video↳ The Moonshots read

Moonshots robotics deep-dive: when machines outnumber staff, name who hands over each task

This Moonshots deep-dive with Figure AI's Brett Adcock covers a campus where robots now outnumber staff, robots working zero-shot in homes they have never seen, and safety designed in from the hardware up. Our read: before a machine takes a physical task in one workflow on Monday, name the person who decides it is ready, what they check first, and how they take the task back.

Organization leads↳ Agent limits & controls

Gartner: 95% of I&O teams had an agent incident, and 15% can stop one automatically

Gartner's survey of 298 infrastructure and operations leaders found 95% had an agent-related incident in the past year and only 15% have automatic circuit breakers, and it advises escalation thresholds that rise with each agent's risk, with high-impact actions kept for human sign-off. On Monday, take one agent in production and draw its line: what it does alone, what goes to a named person, and who can switch it off.

Microsoft and 40-plus firms want every AI workflow to carry its owner and authority

Microsoft launched an AI Transparency Initiative at Ignite with more than 40 organizations, including Accenture, KPMG, Mastercard and Volvo Cars, working toward open standards where each AI workflow carries a record of its purpose, owner, data and delegated authority that can be checked independently of the model provider. On Monday, write that record for one agent-run workflow, one line each for purpose, owner and what it may do without asking, so the check exists before any standard does.

Just Capital: 52% of leaders redefined roles for AI, while 15% made layoffs

Just Capital's quarterly survey finds 52% of corporate leaders redefined or moved roles in the past year while 15% made layoffs, and ServiceNow shows the shape of it: agents now resolve most service-desk tickets, most of that staff moved into new roles and some now supervise the agents. On Monday, pick one workflow an agent is taking on and write down the new role each person on it moves into, including who supervises the agent, before the work changes hands.

TUESDAY · NOV 17

Organization leads

Old rules and incentives, not the tools, are holding AI gains back from profit

In Fortune, board adviser Louisa Loran argues that AI's productivity gains stall before profit because past decisions have hardened into rules and incentives still reward the old way of working. On Monday, list the rules one workflow runs on, ask what each was protecting, keep only the legal, safety and capacity limits fixed, and tie part of the team's target to the shared result.

Uber checks each agent action against the intents its owner allowed

Uber registers every AI agent with a named owner, its permitted intents and its constraints, then checks each proposed action against that record before it runs and keeps a trace that can be replayed when policies change. On Monday, write down for one agent in one workflow who owns it and the few things it is allowed to do, because explicit, per-agent limits are the control Uber leans on most.

Gallup: AI moves people's time to checking work and learning, while meetings stay put

Gallup finds that frequent AI users spend less time gathering information and drafting, about one in five now spend more time reviewing output or learning new tools, and meetings and collaboration have barely moved. On Monday, pick one workflow and decide on purpose where the freed hours go, including who reviews the AI's work and what the new output expectation is, before pressure sets it for you.

MONDAY · NOV 16

Organization leads · video↳ The Moonshots read

Moonshots #311: when a leader steps away, know who makes each call

Moonshots #311 spends a long stretch on Jensen Huang's departure from NVIDIA and who could replace him, with Emad Mostaque joining Peter Diamandis, Salim Ismail and Alexander Wissner-Gross. Our read for Monday: in one workflow, write down who makes each decision when the usual decider is out, and check that the agents in it know where to escalate.

Organization leads↳ Managers, jobs & pay

Cisco pairs its AI power users with senior leaders, the users as mentors

Cisco says its people are twice as likely to use AI when their direct leader does, so it pairs employee AI super users with senior leaders as their mentors and gives every employee four hours a month to learn. On Monday, pair the person who runs one workflow best with AI with the leader who owns that workflow, and let the expert propose the next change to it.

Technology leads↳ Managers, jobs & pay

Security Bank staff have built 3,100 agents, with more human review as stakes rise

Security Bank in the Philippines has trained more than 3,000 employees in agentic AI, and they have built about 3,100 small agents for their own work, while credit, fraud, risk and complaints decisions keep more human judgment and controls. On Monday, let the people in one workflow build the agent for its routine steps, and write down the point where a person still reviews before anything leaves the team.

Charter: leaders trust AI over their own judgment twice as often as staff

Charter's survey of 5,002 U.S. desk workers finds about a third of managers and executives trust AI over their own judgment on work decisions, against 16% of other staff, and leaders are more likely to act on output they doubt. On Monday, pick one leadership decision that leans on AI, such as a budget call, and have it name the colleague with context who checked the output before it goes ahead.

Last week

Nov 9–13 · 19 takes · headlines

The archive

by week · 4 weeks
Week of Nov 29 takesShow
Week of Oct 267 takesShow
Week of Oct 196 takesShow
Week of Oct 122 takesFirst takesShow
The weekly · every Friday

The week, edited down.

Every Friday the best takes of the week go out as one email, with one move to try on Monday. Each issue has its own page. All issues →

Get The H.O.T. Take every Friday.

The week's best takes, one email. Read past issues →

Issue 03 Latest

Rewrite the rules.

The work has already moved; the rules around it have to be rewritten on purpose. With Moonshots #311, five numbers and six more reads.

Read Issue 03 →
Issue 02

Keep the judgment.

The agent can take on the work; the judgment, and the authority to use it, stay with people. With Moonshots #309, and John Longhill and Jani Ashmore on the human layer.

Read Issue 02 →
Issue 01

Name the person.

As agents take on more of the work, the job that grows is deciding who answers for it. The best of the first three weeks, plus John Longhill on why AI is a readiness problem.

Read Issue 01 →
Issue 04 · in the works

Next Friday.

The week's best takes, one email, with one move to try on Monday. Subscribers get it first; it lands here the same day.

The H.O.T. Take is Pegasus Collective's independent commentary. Sources are linked, not reproduced, and the words are ours. References do not imply endorsement. ExO 3.0 frameworks are by Salim Ismail and OpenExO.