ConversationPilot puts one clear coaching card on your screen at exactly the right moment — the next question, an objection response, a suggested follow-up — so you always know your best move.
Updated
Zoom, Teams, Google Meet or any call app · Download for Mac & Windows · Free plan, no card needed · See pricing
Live sales prompts are the visible heart of ConversationPilot: the glanceable coaching cards that appear on your screen during the call, telling you the single most useful thing to do next. Where a recorder shows you what you should have done after the fact, a live prompt shows you what to do now, while the conversation is still in play and the outcome is still open.
Each prompt is a compact card surfaced in the overlay — the next best question to ask, a clean response to the objection that was just raised, a qualification prompt for a gap you still have open, or a suggested follow-up. The card is short enough to absorb without breaking eye contact or losing the thread. It appears about five seconds after the counterpart speaks, sometimes while they are still talking, because a prompt that arrives after the moment has passed is useless, and cards are spaced out rather than constant, so the experience is a colleague whispering in your ear rather than a dashboard demanding to be watched.
It runs as a discreet desktop overlay on Zoom, Teams and Meet, hidden from screen sharing, with no bot in the call. The prompts are suggestive, not coercive — you stay in control and decide whether to take each one — which is exactly what makes them usable on real calls instead of switched off.
A live sales prompt is a real-time coaching card surfaced during the call, condensing the copilot's understanding of the moment into one clear, actionable suggestion. It is the output layer where everything ConversationPilot understands — the transcript, the objections and signals it has picked up, the qualification still to cover — turns into a single move the rep can make.
There are a few kinds. A next-best-question prompt suggests the most useful thing to ask given where the conversation actually is. An objection-response prompt gives a specific reply seconds after an objection lands. A qualification prompt nudges you to cover something the conversation hasn't reached yet — like who signs off, or when their contract renews. And a follow-up prompt suggests the natural next probe on a thread worth pursuing. In every case the card is a suggestion you choose to act on, not a script you are forced to read.
The obvious worry about live prompts is that reading them will pull you out of the conversation. ConversationPilot is built so the opposite is true. Each prompt is a short, glanceable card in a calm overlay: cards come at least eight seconds apart, and each one stays on screen while the next is being written.
The goal is a colleague whispering in your ear, not a screen demanding to be watched. You stay present with the counterpart, glance when a card appears, and absorb it the way you would a nod from a manager sitting beside you. Because the prompts are sparse and well-timed rather than constant, they reduce cognitive load instead of adding to it — you no longer have to hold the entire playbook in your head while also listening, thinking and responding. A prompt that nagged on every sentence would get ignored; one that speaks only when it matters earns its place on the screen.
A live prompt is only live if it arrives before the moment passes. ConversationPilot sends the counterpart's audio in five-second clips, transcribes each one as it closes and writes the card straight away, so a card is on screen about five seconds after the counterpart says something — fast enough to act on before your next sentence rather than after the moment has closed. On a long answer, a new card can arrive while they are still talking.
That speed is a deliberate architectural choice. The live prompts run on Claude Haiku 4.5, a fast model tuned for low latency, while the post-call report runs separately on a stronger model so it never slows the in-call assist. Your microphone and the meeting audio are captured as two separate streams, so the system already knows which words were theirs when it writes the card. Every part of the pipeline is built around the same constraint: get the prompt on screen while it is still useful.
Different moments need different prompts, and ConversationPilot covers the arc of a conversation. Early on, next-best-question prompts push you toward the open questions that surface the real problem rather than a feature dump. When an objection lands, an objection-response card gives you the specific reply. When a qualification topic hasn't come up, a prompt can remind you to ask about budget, authority, timeline or the current solution before time runs out.
When the prospect drops a thread worth pursuing, a follow-up prompt suggests the natural next probe so you do not breeze past genuine interest. And when the prospect finishes a long answer, the next card can hand you an open question, so you have a reason to stop talking and listen. Taken together, the prompts mean you are never alone with a blank moment: whatever the call needs next, the most useful move is one glance away, suited to what was actually just said rather than a fixed script.
Live prompts change the current call, but they are also the fastest way to build durable skill. When a rep is nudged toward better questions, cleaner objection handling and tighter qualification on every conversation, those behaviours become automatic surprisingly quickly. The prompts are training wheels the rep eventually no longer needs.
New hires feel this most. With the playbook surfacing live on their very first call, they practise the right behaviours in real conversations instead of memorising a binder or shadowing for weeks. Over time the prompts become instinct, and the rep internalises the moves. ConversationPilot reinforces this with an automatic post-call report, so the live lesson is echoed in reflection afterward. But the engine of improvement is the constant, in-context repetition that only live prompts provide — being shown the right move, repeatedly, at the exact second it matters.
Generic prompts give generic help. ConversationPilot grounds its live cards in AI Playbooks — Sales Discovery, Enterprise Sales, Customer Success, Investor Pitch and more — each of which tunes the prompts the copilot surfaces. So the cards reflect how your team actually sells rather than imposing a one-size-fits-all script over every call.
Your own material grounds them further. Upload your pitch, price list or methodology to the Knowledge Base and choose which documents ride on each call from the desktop app's document picker, or pick Your brief mode and the coaching follows those documents instead of a preset playbook. The cards are built to stick to what's in your documents and what was said on the call, rather than making up prices, dates or guarantees, and when a figure is missing they hand you the question that gets it. That is what makes a prompt something you can say out loud with confidence.
| Capability | ConversationPilot AI | Recorders / note-takers |
|---|---|---|
| Coaching cards during the call | Live, in about 5 seconds | After-call only |
| Form of guidance | One glanceable card | Long review notes |
| Next best question | Surfaced live | Not provided |
| Objection responses | In the moment | Reviewed later |
| Tuned to your playbook | AI Playbooks and your own documents | Generic |
| Distraction during call | Sparse, well-timed | N/A — nothing live |
Live sales prompts are real-time coaching cards that appear on your screen during the call, condensing the copilot's understanding into one clear move — the next best question, an objection response, a qualification reminder or a suggested follow-up. They appear within about five seconds, while the moment is still live.
It's designed not to. Each prompt is a short, glanceable card, spaced out rather than constant, like a colleague whispering in your ear. Because prompts are sparse and well-timed, they reduce the mental load of holding a whole playbook in your head rather than adding to it.
A card appears about five seconds after the counterpart says something, sometimes while they're still talking. Their audio is sent in five-second clips, and live prompts run on Claude Haiku 4.5, a fast latency-tuned model, while the post-call report runs separately so the in-call assist never slows down.
No. The prompts are suggestive, not coercive. You stay in control of the conversation and decide whether to take each card. That respect for your judgement is exactly what makes the prompts usable on real calls rather than something you switch off.
There are next-best-question prompts, objection-response prompts, qualification prompts that nudge you to cover what hasn't come up yet, and follow-up prompts that suggest the natural next probe. Whatever the call needs next, the most useful move is one glance away.
Yes. AI Playbooks — Sales Discovery, Enterprise Sales, Customer Success and others — tune the prompts to your methodology, and you choose which of your own documents ride on each call, or use Your brief mode so the coaching follows them. The cards are built to stick to what's in your documents and what was said on the call, rather than making up prices or dates.
Real-time prompts, objection handling and qualification — while the call is happening.