Can AI be your assistant? What it can and cannot do for founders

This article is a submission by DonnaPro, a virtual assistant agency based in Slovenia. It places EU-based executive assistants exclusively with CEOs and founders, covering inbox and calendar management, travel planning, research, and operations support across 110+ industries worldwide.

Key takeaways

• AI can already handle much of the mechanical side of executive support: drafting, summarising, transcribing, scheduling and research. For those tasks, it is faster and cheaper than a person.
• It cannot yet decide what matters, manage relationships, read a political subtext or take responsibility when something goes wrong. That judgement is most of what a senior executive assistant does.
• AI has moved the founder’s bottleneck from doing to deciding. Microsoft data shows workers are interrupted every two minutes during core hours, 275 times a day, and faster output means more of it reaches the top.
• The World Economic Forum expects administrative assistant and executive secretary roles to see some of the largest declines by 2030. That decline is in routine admin, not in senior executive support.
• The strongest setup for most founders is a skilled human assistant who uses AI well, backed by a managed team rather than working alone.

The short answer: AI can do the tasks, not the job

AI can be part of your assistant. It cannot yet be the whole of one.

The tools are good and getting better. They draft emails in seconds, summarise long reports, transcribe meetings, suggest meeting times and pull together research that used to take an afternoon. For a founder drowning in admin, it is tempting to conclude that a subscription can replace a person.

The evidence points somewhere more interesting. The World Economic Forum’s Future of Jobs Report 2025 lists administrative assistants and executive secretaries among the roles expected to see the largest decline in absolute numbers by 2030. That finding is real, but it describes routine clerical work: data entry, basic correspondence, simple scheduling. It says far less about the senior end of the role, where the work is judgement rather than output.

At the same time, many leaders report feeling busier with AI, not calmer. Microsoft’s analysis of Microsoft 365 activity found that workers are interrupted every two minutes during core working hours, 275 times a day, and receive an average of 117 emails daily. When everyone around a founder produces more, more of it lands on the founder’s desk.

This article sets out what AI assistants handle well today, where they fall short, and how founders can combine AI with human support to get the capacity AI promised without the overload it often brings.

What AI assistants handle well today

AI is strongest at work that is repetitive, text heavy and easy to check. For these tasks, it is often faster than any person and available at any hour.

• Drafting: first versions of emails, meeting agendas, briefings and social posts, ready to be edited rather than written from scratch
• Summarising: long email threads, reports, contracts and meeting transcripts reduced to the key points
• Transcription and notes: meetings recorded, transcribed and turned into action lists automatically
• Scheduling support: suggesting times across calendars and handling simple booking links
• Research: background on a company, a market or a person before a meeting, in minutes rather than hours
• Formatting and data: tidying spreadsheets, extracting figures from documents, building simple reports

What these tasks have in common is that the right answer is fairly clear, the stakes of a small error are low, and a person can check the result quickly. That last point matters. Every AI output still needs someone to review it, and for many founders, that someone is still them.

Pro tip: Keep a list for one week of every time you correct or rewrite something an AI tool produced for you. The total tells you how much of the time saved you are quietly paying back in review.

Where AI falls short

AI falls short wherever the work depends on judgement, relationships or accountability. These are the parts of executive support that founders value most, and the parts that are hardest to see until they are missing.

Deciding what matters. An AI tool can summarise twenty emails. It cannot reliably decide which three the founder needs to see today, which can wait a week, and which should never reach them at all. That filtering depends on knowing the founder’s priorities, relationships and mood, and those change daily.

Reading context and subtext. A short reply from a long standing investor, a delayed answer from a key client, a board member copying someone new into a thread: each can signal something important. A good assistant notices and flags it. An AI tool treats it as one more message.

Managing relationships. Rescheduling a meeting with a major client is not a calendar problem. It is a relationship problem, and it needs tact, timing and sometimes a phone call. The same goes for handling a difficult supplier or protecting a founder’s time without offending anyone.

Taking responsibility. When a flight is cancelled at midnight or a contract goes to the wrong person, someone has to own the problem and fix it. AI tools do not carry accountability. A person does.

Discretion. A senior executive assistant handles board papers, personal finances, health appointments and confidential negotiations. Founders need to know who can see that information and under which rules it is protected.

Why AI can leave founders more overwhelmed

AI removed the limit on how fast work can be produced, but not the limit on how much one person can decide. The bottleneck has moved from doing to deciding.

Before AI, a competitive analysis took a week, so a founder saw one a quarter. Now it takes minutes, so they may see one a week. The same is true for reports, proposals, options and drafts. Each is useful. Together, they mean more decisions arrive at the top, faster, and all of them look urgent.

The instinctive fix is another tool. It rarely works, because the problem is not a shortage of output. It is the lack of a filter between a growing flow of inputs and the one person at the end of it.

That filter used to be a person: a senior executive assistant or chief of staff who decided what reached the leader and in what form. As companies flattened and software promised to do the job, many founders lost that layer. AI has made its absence much easier to feel.

How the executive assistant role is changing

The executive assistant role is splitting in two. Routine admin (data entry, basic scheduling, simple correspondence) is being absorbed by software, which is what the World Economic Forum’s forecast reflects.

Senior executive support is moving the other way. The assistants in highest demand now use AI to clear the mechanical work in minutes, then spend their time on what AI cannot do: filtering, preparing decisions rather than just data, and managing the relationships around the founder. An assistant who works this way can cover a scope that used to need two people.

For founders, this changes the hiring question. It is no longer “AI or an assistant?” It is “which assistant knows how to get the most out of AI?”

Pro tip: When interviewing an assistant or a provider, ask for specific examples of work they have automated and what they did with the time it freed up. General familiarity with AI tools is not the same as using them in daily work.

How to combine AI and a human assistant

The most effective setup gives AI the volume and gives the human assistant the judgement. The assistant sits between the tools and the founder, using AI to process the work and deciding what reaches the founder.

In practice, that looks like this:

• AI processes the inbox. Tools sort, tag and summarise incoming mail and draft replies to routine messages.
• The assistant filters. They approve or edit the drafts, handle what they can alone, and pass the founder a short daily summary of the few items that need a decision.
• AI prepares the material. Meeting transcripts, research and first drafts of briefings are generated automatically.
• The assistant prepares the decision. They turn that material into a one page brief with the context, the options and a recommendation, so the founder decides rather than reads.
• The assistant owns the follow through. Actions from meetings are tracked and chased until they are closed, with AI handling reminders and the assistant handling people.

The result is that the founder sees less, not more, and what they do see is ready for a decision.

Pro tip: Agree on one channel through which everything reaches you, such as a single daily summary from your assistant. If AI tools and team members can all send you material directly, the filter does not exist.

Questions to ask before choosing AI enabled executive support

• How does the assistant use AI in daily work, and which tools do they use?
• Who checks AI generated work before it reaches me or my contacts?
• Which of my data goes into AI tools, and where is it stored and processed?
• Which data protection rules apply to the assistant and the tools they use?
• How is the assistant trained as AI tools change?
• What happens to my workflows and knowledge if the assistant leaves?

What good AI-enabled executive support looks like

The most reliable version of this setup is a dedicated human assistant, trained in AI tools from the start and backed by a managed team, so the quality of the work does not depend on one person and continuity does not depend on that person staying.

Working in overlapping time zones and under a clear data protection regime matters too, since much of the work involves confidential material.

Founders who want the capacity AI promised without the overload can see how a dedicated virtual executive assistant works or book a strategy call to plan the handover.

Sources

• The Future of Jobs Report 2025, World Economic Forum, January 2025: administrative assistants and executive secretaries among the largest declines by 2030
• Breaking down the infinite workday, Microsoft WorkLab, June 2025: 275 interruptions a day, 117 emails a day, based on Microsoft 365 data to February 2025
• AI Gave Every CEO More Capacity. Why Are They More Overwhelmed Than Ever?, Andrew Palmer, CEO Today, June 2026: the shift from doing to deciding, the missing filter