Voice Q&A for managers: getting an instant report in the meeting

On Monday morning, while the management meeting is under way, someone asks about the sales target. Another says “I’ll look it up and tell you”, a laptop is opened, people click between screens and the subject moves on to the next item. The figure often arrives by e-mail after the meeting is over; the decision has long since been made. Voice Q&A for managers means asking questions about company data by voice during a meeting, without hunting for a report screen, and getting a short reply and a screen built for the question straight away.

In this article we explain how Fero, the YazSis Enterprise AI assistant, can be used in a meeting, through a demo flow prepared with the product’s sample data.

One meeting, four questions

1. “How are our sales today?” Together with a short spoken reply, the daily sales summary appears on screen: the day’s sales, progress toward the daily target, number of orders and the difference from the same day last week.
2. “Which sales reps are below target?” The question takes the same subject to a sales rep breakdown. The screen is not reset; those below target come to the front.
3. “What is this rep’s problem?” The customer breakdown opens. Because in the sample data the whole drop comes from two customers, the meeting turns toward a concrete decision about two customers instead of a general discussion.
4. “Is there anything else I should pay attention to today?” The executive summary arrives: late loadings, critical stock and a customer below budget. Here Fero itself brings forward what was not asked but matters.

The four questions were answered with one-sentence questions, without navigating a report menu. The tempo of the meeting is not broken, because instead of waiting for the answer the conversation can go on. In the demo interface the assistant’s state is also visible: listening, analysing, answering and ready. The response time measured in the demo is about two seconds; in a real environment the time varies with the data source and the scope of the question.

This flow means moving report preparation from before the meeting into the meeting itself. In the classic arrangement an analyst guesses the likely questions before the meeting and prepares slides; pages nobody asks about go to waste, while questions that are asked but were not prepared are left hanging. In voice Q&A the meeting itself sets the agenda: managers ask for whatever figure they need at that moment, and the answer is built around the question.

The most expensive moment in a meeting is when the answer is postponed with “we’ll look into it later”.

The advantage and the limit of asking by voice

• Advantage, no typing: Instead of leaning over a keyboard, the flow of the meeting is kept and everyone looks at the same screen.
• Advantage, follow-up questions: Short follow-ups such as “show the top five” or “look at the last three months” keep the context; there is no need to explain the subject from scratch each time.
• Limit, ambient noise: In a crowded meeting room, if several people talk at once, picking up the speech can become difficult. It is a good habit for one person to ask the question and for the meeting to fall quiet until the answer comes; a short silence is cheaper than the cost of a misunderstood question.
• Limit, confirming the figures: Before an important decision, the figure that was spoken should be read again against the indicator on screen.

What to expect in the first meetings?

In their first attempts managers mostly ask questions that are too broad: a question like “How is the company doing?” leaves open which period and which measure you mean. Short, single-subject questions that state a period bring a clearer answer: something like “Where are we against this month’s sales target?” After a few meetings the team learns which questions work, and the question list matures by itself. Planning this learning period from the start prevents small hitches in the first meeting from damaging trust in the assistant.

A preparation list before the meeting

• Question list: Write down in advance the five or six questions that are always asked in the meeting and try them once with the assistant.
• Data source: The source that is ready today is MySQL. If your data is in another system (such as SQL Server, PostgreSQL, Logo, SAP, Odoo or Excel), it is handled within a project; make clear before the meeting whether it is ready.
• Data freshness: If the target and budget tables are not up to date, a quick answer becomes a quick mistake.
• Fallback: If voice cannot be used, it is possible to ask the question in writing; plan the two together.

The accuracy question: the voice is fluent, the figure is the system’s

Because a spoken reply is fluent, it sounds correct; that is why the real basis for trust should be not the voice but the method behind it. In Fero the AI does not write SQL, data is reached only through defined, read-only business tools, and the system calculates KPIs, percentages and differences. The model cannot state a number that is not in the tool result. This structure makes the figure heard in the meeting a calculated result, not a guess. Even so, it is sensible to compare a figure once with your own records before deciding on it. In the first meetings, placing a few of the assistant’s answers side by side with reports you know is the most practical way to test both trust and data quality; if a difference appears, its cause can be examined together after the meeting.

To try your meeting questions with your own data, you can request a live demo on the Enterprise AI page.