What an AI Sales Assistant Actually Does During a Live Call
Every AI sales tool claims to “coach reps in real time.” Most of them mean they analyze the call after it ends and give the rep feedback hours or days later. That is post-call analysis, not real-time coaching. Real-time AI coaching means the assistant is listening to the live conversation as it happens and delivering prompts, reminders, and information to the rep during the call while they can still act on it.
This blog walks through exactly what happens during a live sales call when Revenue.io Moments is running. What the rep sees on screen. What triggers each prompt. What information appears and when. And what data is generated by the time the call ends. No marketing language. Just the actual in-call experience from the rep’s perspective.
Before the Call Starts
When the rep clicks to dial from a Salesforce record, the AI assistant activates automatically. No separate app to open. No toggle to enable. The assistant runs inside the same Salesforce window the rep is already working in.
Before the prospect answers, the assistant loads context from the Opportunity record: deal stage, days in current stage, last meaningful activity date, key contacts and their roles, and the coaching score from the rep’s most recent call on this deal. If the rep scored below 60% on a specific methodology criterion last time (for example, they did not ask about the decision process), that criterion is flagged as a priority for this conversation.
The rep does not need to review this context manually. It is surfaced automatically in a compact panel alongside the dialer. By the time the prospect says “hello,” the rep has the deal context and the coaching priority without having scrolled through a single record.
During the Conversation: What Triggers Prompts
The AI assistant listens to the live conversation and triggers prompts based on specific events in the dialogue. Prompts are not random. Each one fires because something happened (or did not happen) in the conversation that the assistant was trained to detect.
Trigger 1: Competitor Mention
The prospect says “we are also looking at Gong” or “your competitor quoted us $X” or references any competitor by name. The assistant detects the mention and surfaces a competitive battlecard on the rep’s screen within 2 to 3 seconds.
The battlecard contains the specific talking points for that competitor: where Revenue.io is stronger, where the competitor is weaker, common objections prospects raise when comparing, and the reframe language that experienced reps use to redirect the conversation. The rep does not need to memorize competitive positioning for every competitor. The information appears at the exact moment it is needed.
This is one of the highest-value triggers because competitive conversations are high-stakes and time-sensitive. A rep who fumbles a competitor comparison for 30 seconds while trying to remember the positioning may lose the frame for the rest of the call. A rep who glances at the battlecard and responds confidently within 5 seconds maintains control of the conversation.
Trigger 2: Pricing Discussion
The prospect asks “how much does this cost?” or “what is the pricing?” or “can you send a proposal?” The assistant detects the pricing trigger and surfaces the configured response guidance: whether to discuss pricing on this call or defer to a formal proposal, the approved discount thresholds by deal size, and the value reframe language that redirects the conversation from price to business impact before quoting numbers.
This trigger prevents two common pricing mistakes: quoting before value is established (which anchors the conversation on cost rather than impact) and offering unauthorized discounts under pressure (which erodes margins). The prompt reminds the rep of the process in the moment rather than relying on them to remember it under the pressure of a live pricing question.
Trigger 3: Methodology Criteria Not Covered
This is the trigger that drives methodology adherence. The assistant tracks which methodology criteria (MEDDIC, BANT, Challenger, or custom framework) the rep has covered during the conversation and which are still outstanding.
If the call reaches the 15-minute mark and the rep has not asked about the decision process, a prompt surfaces: “Decision process not yet discussed. Consider asking: ‘Walk me through how your team evaluates and approves a solution like this.'” If the economic buyer has not been mentioned by the 20-minute mark, a prompt appears: “Economic buyer not identified. Consider asking: ‘Who on your side owns the budget for this initiative?'”
The timing thresholds are configurable by call type. Discovery calls may trigger methodology prompts earlier because the entire purpose is information gathering. Demo calls may trigger them later because the first portion is presentation. The assistant adapts to the call type rather than applying the same timing rules to every conversation.
This trigger is what turns methodology from something reps learn in training and forget within weeks into something they execute on every call because the system reminds them in the moment. Deals that die in mid-pipeline almost always show a pattern where reps stopped covering methodology criteria because they felt comfortable with the deal. The prompt prevents that comfort from becoming a gap.
Trigger 4: Objection Detection
The prospect says “I am not sure this is the right time” or “we tried something like this before and it did not work” or “I need to think about it.” The assistant detects common objection patterns and surfaces the recommended response framework.
The response guidance is not a script. It is a framework: acknowledge the concern, ask a clarifying question to understand the specific issue, then address the underlying concern with the relevant positioning. For the “tried it before” objection, the prompt might suggest: “Acknowledge their experience. Ask what specifically did not work. Position the difference between their previous solution and your approach.”
Experienced reps handle objections instinctively. New reps freeze. The objection trigger is most valuable for the reps who need it most: new hires in their first 90 days who have not yet developed the instinctive responses that come with experience. For these reps, the prompt is the difference between a stumbled response that kills the deal and a confident response that advances it.
Trigger 5: Talk Ratio Alert
If the rep has been talking for more than 70% of the conversation at any point past the 5-minute mark, a gentle prompt surfaces: “The prospect has spoken for less than 30% of the conversation. Consider asking an open-ended question.” This is the simplest trigger and one of the most effective because talk ratio is the most reliable leading indicator of call quality. Reps who talk too much on discovery calls miss the information they need. Reps who let the prospect talk uncover the insights that advance deals.
The threshold is configurable. Demo calls naturally have a higher rep talk ratio than discovery calls. The assistant adjusts by call type so the alert only fires when the talk ratio is outside the expected range for that specific conversation type.
Trigger 6: Key Moment Bookmarking
When the prospect says something significant (names a budget figure, mentions a timeline, references a competing solution, states a specific pain point, or commits to a next step), the assistant automatically bookmarks that moment in the recording with a timestamp and a label. The rep does not press a button. The assistant detects the significance and marks it.
After the call, the manager can jump directly to bookmarked moments rather than listening to the entire recording. A 45-minute call with 6 bookmarked moments can be reviewed in 5 minutes because the manager goes directly to the moments that matter. This is what makes coaching from call data practical at scale rather than requiring managers to listen to full recordings.
What the Rep Experiences
The most common concern about real-time prompts is that they will be distracting. In practice, the opposite is true. The prompts are designed to be glanceable: a short text card that appears in the rep’s peripheral vision and can be read in 2 to 3 seconds without breaking eye contact with the screen (for video calls) or losing focus on the conversation (for phone calls).
Reps report that the prompts function like a copilot rather than a distraction. They glance when they need to (competitive mention, pricing question) and ignore prompts they do not need because they are already covering the criterion. The assistant does not interrupt the conversation. It surfaces information alongside it.
New reps rely on prompts more heavily during their first 60 to 90 days. Experienced reps glance at them occasionally and find them most useful on competitive and pricing triggers where the specific data points (competitor pricing, approved discount thresholds, battlecard details) are hard to memorize across multiple competitors. Over time, most reps internalize the methodology prompts and the assistant becomes primarily a competitive reference and deal context tool rather than a methodology reminder.
After the Call Ends
When the call ends, the assistant generates several outputs automatically. The rep does not log anything manually.
AI-generated call summary. A structured summary of the conversation including topics discussed, objections raised, commitments made, and next steps agreed upon. The summary is written to the Opportunity record in Salesforce within minutes. Conversation intelligence processes the full transcript while the summary provides the quick-reference version.
Methodology coaching score. An AI-generated scorecard evaluating which methodology criteria the rep covered and which they missed, scored against the framework configured for this call type with stage-appropriate weighting. The score is written to a custom field on the Opportunity record so it can be tracked over time and correlated with deal outcomes.
Activity logged automatically. The call duration, participants, recording, transcript, summary, and coaching score are all logged to the correct Salesforce records automatically. No manual entry. No CRM update the rep needs to do before moving to the next call. The CRM is complete because the AI handled it.
Bookmarked moments available for coaching. The key moments bookmarked during the call are tagged with timestamps and labels in the recording. The manager can review them in the coaching dashboard without scanning the full call.
Next-best-action updated. Based on what happened on the call (new stakeholders mentioned, objections raised, competitive evaluation confirmed, next steps agreed upon), the guided selling system updates the recommended next action for this deal. When the rep finishes the call and returns to their daily action queue, the queue already reflects what they learned in the conversation they just had.
What This Means for Managers
The AI assistant changes the manager’s job from finding coaching opportunities to acting on them. Before real-time AI coaching, a manager who wanted to coach a rep on a specific deal had to listen to the recording (30 to 45 minutes), identify the coaching moment (variable), and schedule a conversation to discuss it (days later). By the time feedback was delivered, the rep had moved on to other deals and the coaching moment had lost its immediacy.
With the AI assistant, the manager receives bookmarked moments (5-minute review), a methodology score (instant), and a summary (2-minute read) within minutes of the call ending. The coaching conversation can happen the same day while the call is still fresh in the rep’s memory. And the methodology score tells the manager exactly which criterion to coach without requiring them to diagnose the gap themselves.
Frequently Asked Questions
Are real-time prompts distracting during calls?
No. Prompts are designed to be glanceable: short text cards that appear in the rep’s peripheral vision and can be read in 2 to 3 seconds. Reps glance when they need to and ignore prompts they do not need. New reps rely on prompts more heavily during their first 60 to 90 days and experienced reps use them primarily for competitive and pricing triggers where specific data points are hard to memorize.
How does the AI know when to trigger a prompt?
Prompts are triggered by specific events in the conversation: keyword detection (competitor names, pricing language, objection phrases), methodology timing (criteria not covered by a configurable time threshold), conversation dynamics (talk ratio exceeding threshold), and significant moments (budget mentions, timeline commitments, next step agreements). Each trigger is configured by your team and can be adjusted based on call type and deal stage.
Does the AI tell the rep what to say?
No. The AI surfaces frameworks, battlecards, and reminders. It does not generate scripts or specific sentences for the rep to read. The competitive battlecard provides talking points and positioning, not a word-for-word script. The methodology prompt reminds the rep which criterion to cover and suggests a question, but the rep phrases it naturally in their own words. The AI is a copilot, not an autopilot.
What happens to the data after the call?
The call recording, transcript, AI summary, methodology coaching score, and bookmarked moments are all written to the Salesforce Opportunity record automatically. Everything lives natively in Salesforce, respects your existing permissions and sharing rules, and is available for coaching, pipeline review, and reporting without accessing a separate platform.
Does this work for in-person meetings too?
Revenue.io Mobile supports in-person meeting recording with one tap. The conversation is recorded, transcribed, and scored through the same AI coaching system. Post-call outputs (summary, score, activity log) are identical to phone calls. Real-time prompts during in-person meetings are delivered through the mobile app. For the full guide on recording in-person sales meetings including the legal framework by state, see our in-person recording guide.
Conclusion
An AI sales assistant during a live call is not a gimmick and it is not a distraction. It is a system that surfaces the right information at the right moment: competitive battlecards when competitors are mentioned, pricing guidance when pricing comes up, methodology reminders when criteria are missed, and objection frameworks when the prospect pushes back. After the call, it generates the summary, the score, the activity log, and the bookmarked coaching moments without the rep typing a single word.
The result is a rep who is better prepared during the conversation, a manager who can coach from specific moments rather than general impressions, and a CRM that is always complete because the AI handled the documentation. That is what an AI sales assistant actually does during a live call. Not in a demo. On every call, every day, for every rep.