How an executive uses AI day to day: emails, reports, meetings and decisions

The gain shows up when AI has a fixed place in your week, tied to tasks you already do: Monday’s difficult email, the monthly report, the board meeting, the proposal waiting for a reply. Used only when there’s time to spare, it becomes a curiosity. Tied to those moments, it becomes a habit, and habit is what separates people who have tried AI from people who work with it. None of this requires knowing how to code.

How to use AI as a personal assistant

The best way to think about AI is as a very fast chief of staff who just arrived and knows nothing about the company. They deliver well when given context and deliver poorly when given a one-line order.

Before asking, say five things, in plain running text:

• Context: the situation, what has already happened and what your role in it is.
• Audience: who will read or hear it, and what that person already knows.
• Goal: what needs to happen next, whether an approval, a decision or a change in behavior.
• Tone: firm, warm, technical, informal.
• Format: a short email, a one-page memo, talking points for a meeting.

Attach the supporting material (the report, the previous minutes, the proposal) instead of summarizing it from memory. And treat the first answer as a draft: ask for adjustments the way you would with someone on your team.

Emails, memos and announcements

Writing is the use with the fastest return, because executives write all day and almost always in a hurry. The gain lies in getting to the second draft in minutes, with you deciding what stays.

An example: you need to tell the sales team about a change in the discount policy that no one will like. Describe the change, the reason, who will be affected and the tone you want, direct and without apologizing, and ask for a short email. Then flip the roles: ask AI to read the text as a skeptical manager and point out what will cause friction or be misread.

The same goes for the memo to the board and the reply to a supplier who delivered late. Two review requests work on almost any text: cut it in half without losing information, and list the sentences that could be read two ways.

Reports and spreadsheets without waiting for BI

AI doesn’t replace BI, but it answers the question that isn’t on the dashboard, when you need it. That’s where the time spent on analyses and reports drops the most.

An example: the monthly results report arrives the day before the budget meeting. Attach this month’s report and last month’s and ask what changed significantly, which numbers break from the trend and which questions you should ask the CFO. With the expense spreadsheet, ask for the biggest variances from budget, by area, pointing to the row each number came from.

Always check the numbers against the source. AI misreads badly formatted tables, swaps columns and sometimes calculates wrong with complete confidence. Ask it to point to the page or cell of each number it cites, recalculate the ones that will support your decision and never take a number you haven’t checked to the board.

Meetings: before, during and after

AI helps in all three phases of a meeting, and preparation is usually the most neglected.

• Before: with the previous minutes, the emails on the topic and the material sent, ask for a one-page briefing and an agenda with time per item.
• Before the board meeting: ask AI to play a skeptical director and list the toughest questions your deck will provoke. Answer them before the meeting.
• During: if company policy allows it and everyone is aware, the transcript frees you to lead instead of taking notes.
• After: from the transcript or your notes, ask for decisions, owners, deadlines and the follow-up email. Review before sending, because AI sometimes records as decided what was only discussed.

Decisions: devil’s advocate, scenarios and research

In decisions, AI is more useful for testing your reasoning than for giving the answer. It tends to agree with whoever is asking, so explicitly ask it to disagree.

• Devil’s advocate: before approving a supplier’s proposal or opening a new location, present your recommendation and ask for the strongest arguments against it.
• Scenarios: describe the decision and the assumptions and ask for scenarios, with what would need to be true in each one. The value lies in the assumptions that get exposed, not in the forecast.
• Deep research: deep research tools consult many sources and deliver a report with references. Use them to map regulation, competitors or a new industry, and open the sources that support the main conclusions.

A weekly routine

The routine that works is small and tied to moments already on the calendar:

• Monday morning: with the week’s calendar and open items, ask for the priorities, what can be delegated and what needs preparation.
• Before each important meeting: a one-page briefing and the tough questions.
• On report day: an assisted reading, with questions for the responsible area and numbers checked.
• Friday afternoon: a draft of the weekly note to the team and the list of what’s still open.

Almai starts training with the board, and the first stage of the AI-First Executive Journey is exactly this: AI in the routine of the people who decide. Those who decide need to feel that gain in their own work before asking the same of their team. For founders and C-level executives who prefer a one-on-one format, the Exclusive 1:1 Classes start from this same routine.

AI’s first ROI is in the productivity of the people who decide.

What not to delegate

AI can prepare almost everything, but some things remain yours:

• The decision: AI puts together options and arguments, but whoever signs is accountable.
• The final numbers: nothing goes to the board, the bank or the market without checking it against the source.
• Difficult conversations: a dismissal or tough feedback can be prepared with AI, but you lead them, in your own words.
• Judgment about people: evaluation, promotion and hiring call for human judgment and care with personal data.
• Critical review: text that looks finished before it’s any good is workslop, and when it comes from the executive, the team understands that this is the standard. It’s worth setting quality standards for AI-assisted work, starting with your own.

On data, the practical rule is to use the enterprise tool your company has approved, never a personal account. The major vendors’ enterprise plans usually state in the contract that data isn’t used to train models; check your tool’s contract. Sensitive personal data, such as health or biometric data, calls for extra care under data protection law (in Brazil, the LGPD). The guide on how to use AI without exposing confidential data details what’s allowed and what isn’t.

How to start this week

1. Confirm which AI tool your company has approved and log in with your enterprise account.
2. Choose a difficult email or memo from this week and draft it with AI, describing the context, audience, goal, tone and format.
3. Before sending, ask for a critique of your own text.
4. Take the next report to AI and come away with questions for the area, checking every number you use.
5. Prepare the next important meeting with a briefing and the list of tough questions.
6. On Friday, note what worked and save the requests that paid off to reuse the following week.