AI for client meetings (consulting)

A senior consultant walks into a discovery call with a CFO. The CFO opens with a number from page 14 of the diagnostic, asks why the working-capital assumption moved 80 basis points between versions, and waits. The consultant knows the answer. It is somewhere between a Slack thread, a partner's voice memo, and a model tab they were staring at on Tuesday. The silence stretches. The CFO checks their watch.

This is the part of consulting nobody puts in the pitch deck. The work is mostly recall under pressure. You have read the data room, you have built the model, you have rehearsed the framing with the partner. Then the client asks the one question that is two questions to the left of what you prepared for, and you are alone with your memory in front of someone paying $4,000 a day for it.

Whisply was built for that exact second. It listens to the call through your Mac microphone, watches the documents and dashboards open on your screen, and surfaces the figure, the precedent, or the on-brand sentence you need before the silence becomes a problem. The client hears you, not the assistant. The assistant is for you alone.

You open the meeting in Zoom or Teams as you normally would. Whisply is already in your menu bar, a small icon next to the battery. The client starts talking. You listen. When they ask the question that needs a precise answer, you tap Cmd+Return and Whisply opens a thin overlay on top of your screen, only visible to you.

Type the question or just let it read the live transcript of the meeting. It pulls from what it has seen on your screen in the last hour, the proposal PDF you had open, the Excel model in the next tab, the partner's email thread in Outlook. It writes back a two-line answer with the figure and the source. You read it, you say it in your own words, you keep eye contact with the camera.

When the client shares their screen and asks you to share yours, Whisply does not appear in your shared window. The overlay sits in a system layer that screen sharing in Zoom, Teams, Webex, Meet, and Loom does not capture. You can keep the assistant open while you walk the client through your slides. They see the deck. You see the deck plus the prompt that reminds you which slide the partner wanted you to skip.

Most firms have tried to build their own version of this. A partner gets excited about an internal search tool, a few engagement teams pilot a vector database stuffed with old decks, the rollout stalls when someone realizes the IP exposure of pushing client material into a third-party index. Two quarters later the budget is reallocated and nobody mentions it again.

Whisply skips the build. It does not require you to upload your knowledge base anywhere. It reads what is on your screen during the call, which for a working consultant is almost always the document that matters. The proposal is open. The model is open. The benchmarking deck is open. Whisply sees what you see, then adds the language model on top.

Models are included in the subscription. No API key, no separate vendor invoice, no procurement memo to sign. The Pro plan at $19.99 a month removes the daily message cap that the Free tier uses, which is usually the version a working consultant lands on within a week of trying it.

Every consulting firm has a house style. McKinsey writes in pyramids. BCG starts with the so-what. Bain repeats the answer three times. Boutiques have their own grammar, their own forbidden words, their own preferred structure for an executive recommendation. The junior consultants who get promoted are the ones who learn that grammar fastest, and the ones who stall are the ones who keep writing like they did in business school.

Whisply learns the framing by reading the documents you actually use. Drop a partner's edited memo into your open windows, and the overlay will start to mirror the structure when it drafts a recommendation back to you mid-call. Ask it to rewrite a client objection in the firm's preferred answer-first form, and it will give you the sentence in a beat. You can paste it into a follow-up email the moment the call ends, while the client is still on the line wrapping up.

This is the part the recap tools miss. A summary that arrives tomorrow does not help you write the answer today. The answer-today version is what clients pay for, and the assistant has to be in the room with you to do it.

Clients say wrong things in meetings. They quote a stat from a McKinsey report they half-read, they misremember their own Q3 number by a factor of ten, they attribute a competitor move to the wrong quarter. The instinct of a junior consultant is to nod and move on. The instinct of a senior consultant is to push back politely with the correct number, which requires having the correct number.

Whisply runs a quiet fact check in the background. When the client cites a figure, you can ask the overlay to verify it against the materials you have open, or against the model's general knowledge. The answer comes back in a few seconds with a confidence note. If the client is wrong, you decide whether to correct them in the call or save it for the follow-up.

This matters more in pre-sale than in delivery. The pitch where you politely correct the prospective client on their own market share is the pitch that gets the work. They walk away thinking the team across the table is sharper than the team they have now. That is the whole game.

The client sees a consultant on a video call, looking at the camera, answering questions in real time, occasionally glancing down to check a note. That is it. They do not see a second participant. They do not see a bot named Notetaker. They do not see a recording indicator. They do not see the Whisply overlay because it does not get captured by Zoom, Teams, Webex, Meet, or Loom screen sharing or by the cloud recording.

This is not a marketing claim, it is a system-level behavior on macOS. The overlay sits in a layer that Apple's screen capture APIs do not return to the recording process. We explain the full mechanism on the undetectability page. The short version is that the consultant gets to use a private tool privately, the same way they would use a notepad next to the laptop, except the notepad answers questions.

There is no recording of the client made by Whisply. The audio is processed for the meeting and not archived as a transcript anyone else can see. If the client asks whether you are using AI, you answer them honestly. The product is built for the consultant who wants to be more capable in the moment, not for the consultant who wants a paper trail of every word their client said.