AI agents for CEOs: 6 ways to win back time to think

Most leaders do not lack information. They lack the time to think about it. The week fills with status meetings, long email threads and documents to find before each decision. On 30 October 2026, the Financial Times reported that CEOs now use AI agents to get that time back. Satya Nadella uses custom agents at Microsoft. Brian Chesky says an information system at Airbnb cut his meetings by about half.

We agree with the article's main point. The value of an AI agent is not the number of tasks it does. It is the amount of leadership attention it gives back. Here are six ways to get that attention back, and how we set them up in d5s.

• A morning brief replaces the first hour of searching across email, Microsoft Teams and the CRM.
• A company knowledge agent answers your questions without a request to an analyst.
• Meeting agents prepare each call and write down the decisions after it.
• A thinking partner tests your plan before your team or your board does.

1. The return of an AI agent is attention, not tasks

Measure an agent by the hours of focus it gives back to you, not by the tasks it completes.

A leader's time has a high value because it goes to decisions. An agent that drafts 50 emails can still add to your day if you must check each one. An agent that removes one weekly status meeting gives back a full hour for planning.

So start from your calendar, not from a list of tasks. Find the hours that you spend to collect information, to prepare for meetings and to chase follow-ups. These are the hours that an agent can take over. Judgment, relationships and priorities stay with you.

2. A morning brief replaces the first hour of searching

A scheduled agent reads your approved sources before you start and gives you only what needs a decision.

A morning brief is the easiest first job for a leader's agent. The agent reads email, calendar, Microsoft Teams, the CRM and the news that you select. Then it sorts what changed into three groups: what you must decide, what you must know and what you must prepare for.

A good brief links every item to its source. It also says when a source could not be checked, so "nothing changed" never hides a failed read. In d5s, you can set up a morning brief in minutes. You choose the days and the time it runs, the sources it reads and what it puts first.

3. An agent finds answers across the company, so you do not wait for an analyst

A knowledge agent searches your connected tools and gives you a sourced answer in minutes, not days.

Many decisions wait for one fact. What did we promise this customer? Why did churn go up in the north region? Which teams asked for more headcount? A leader usually sends the question to an analyst or a manager, and the answer comes back days later.

An agent can answer many of these questions directly. It searches email, Microsoft Teams channels, shared files, the CRM and your other connected tools. Then it writes a short answer and links each point to the message, document or record that it came from. When it cannot find an answer, it says so and tells you where it looked.

This is close to the information system that Chesky describes. The difference is that you do not need an internal team to build it. The agent reads only the connectors that you give it, so it does not show data that you did not approve.

4. Agents prepare your meetings and write down what you decided

An agent can join Google Meet, Microsoft Teams and Zoom calls, take notes, and bring a brief before the next one.

Many meetings exist only to share status. A brief before the call removes some of them. For the rest, an agent can make each meeting shorter and more useful.

Before the call, the agent collects what you agreed last time, the open questions and the recent changes for each attendee. During the call, a d5s agent joins as a listener. It writes live notes that separate decisions and commitments from ideas and open questions. After the call, it can draft the follow-up email. The email waits for your approval before it goes out.

This is the shift that the FT article describes. The information moves to you, so you do not have to go to a meeting to get it.

5. A good agent challenges your plan

Use an agent as a thinking partner: ask it to argue against your plan, compare options and list the risks.

The FT article notes that leaders use AI to test ideas and improve their writing, not only to make content. An agent that knows your company context gives better challenges than a general chatbot. It can read last quarter's numbers, the customer feedback and the board pack before it replies.

Try these requests with your agent:

• "Argue against this hiring plan. Use our pipeline numbers."
• "Compare these two price options for the renewal. List what each one risks."
• "Read this board memo as a skeptical investor. What questions will they ask?"

You can also rehearse a board meeting before the real one. In a d5s project, create one agent for each board member, for example the lead investor, the finance expert and the independent director. Give each one its own instructions, its own priorities and, if you want, its own model. Then share your board memo and ask each agent in turn, with an @ mention, for its hardest questions. You go into the real meeting with your answers ready.

The answers are a starting point for your judgment, not a replacement for it.

6. You stay in control: agents ask before they act

Reading is allowed by default, and anything that sends or changes data waits for your approval.

The FT article also names the limits. AI can be wrong, and trust and judgment stay with people. An agent for a leader must be built for that.

In d5s, you set each action to Allow, Ask or Deny. By default, an agent can read, but it asks before it sends an email, changes a CRM record or posts a message. Sources in the brief let you check a claim in one click. Each agent also gets only the connectors you give it, so a board brief cannot read data that you did not approve. See how we handle security.

How to start with an AI agent as a CEO

Give your first agent one recurring job whose result you can check, such as a morning brief.

1. Choose one job that takes your time each week. A morning brief or meeting preparation are good first choices.
2. Connect only the sources that the job needs.
3. Read the first results and correct the instructions.
4. Add a schedule when you trust the output.
5. Add the next job, such as meeting notes or follow-ups.

A personal agent is a ready-made starting point. It does the preparation and follow-up work. You keep the decisions.

Questions and answers