There is a moment in every discovery call that determines whether the deal has a future. It usually happens in the first five minutes. The prospect says something — a throwaway comment about a recent reorg, a frustration with a current vendor, a budget pressure they hadn’t planned to mention, and the rep either catches it and follows the thread, or keeps moving through their script and loses it forever.
That moment has nothing to do with AI. It never will. It is pure human attention, and no tool is going to replace it.
But everything that happens before that moment, and everything that happens after it, is where AI is quietly rewriting the rules of how discovery calls work.
The prep that used to take 45 minutes now takes five. The notes that used to get typed up in a rush between calls now get generated automatically. The follow-up email that used to sit in drafts until tomorrow morning now lands in the prospect’s inbox before they’ve closed their laptop. And the rep who used to have to choose between listening and note-taking can now do both, because AI is handling one of them.
That’s what this post covers. The before, the during, and the after of a discovery call, and where AI fits into each.
Before the Call: Research That Actually Gets Done
Ask any sales manager what their reps should do before a discovery call, and they’ll give you the right answer: research the company, review the prospect’s LinkedIn, check recent news, understand the competitive landscape, and review any prior touchpoints in the CRM. (refresher on finding the right prospects)
Ask those same reps how often they actually do all of that before a call, and you’ll get a more honest answer.
The problem isn’t laziness. It’s time. A rep running eight discovery calls a week, plus follow-up, plus pipeline management, plus internal meetings, doesn’t have 45 minutes of focused research time per call. They have five minutes, maybe ten, and they use it to skim the company website and hope for the best.
AI doesn’t solve the time problem by making reps faster researchers. It solves it by doing the research for them.
Here’s what a five-minute AI-assisted pre-call brief can include:
- Company overview: size, revenue range, recent growth or contraction, key markets.
- Recent news: funding rounds, acquisitions, leadership changes, product launches, press mentions.
- The prospect’s LinkedIn activity: recent posts, career history, shared connections, topics they engage with.
- Known pain points based on their industry and company stage.
- Prior conversation history pulled from your CRM, summarized in plain language.
- Competitive context: what tools they’re likely using and where the gaps tend to be.
Tools like Clay, Apollo, and even a well-prompted ChatGPT session can pull most of this together in minutes. Some CRMs now generate pre-call briefs automatically when a meeting is on your calendar.
The rep who walks into a discovery call with that brief in hand isn’t just better prepared. They’re more confident, more focused, and more likely to catch the signal when it appears because they’re not spending mental energy on basic orientation.
Preparation isn’t what earns trust on a discovery call. But the absence of it destroys it faster than almost anything else.
During the Call: What Real-Time AI Can and Can’t Do
This is where the conversation about AI and discovery calls gets complicated, and where it’s worth being direct about what’s realistic.
Real-time AI assist tools have gotten genuinely useful. Platforms like Gong, Chorus, Fireflies, and Otter transcribe your calls as they happen, flag key topics, surface suggested questions, and track talk time ratios in real time. Some will prompt you with battle cards when a competitor is mentioned. Some will alert you when a pricing objection comes up, or when a prospect uses language that signals urgency.
That’s real capability. It’s also easy to overuse.
The rep who’s watching an AI dashboard during a discovery call instead of watching the prospect’s face, or listening to the hesitation in their voice, or noticing that they changed the subject when a certain topic came up, is making a trade that doesn’t work in their favor. The tools are useful as a safety net and a coach. They are not useful as a replacement for presence.
Here’s the practical guideline I’ve landed on after watching a lot of reps use these tools:
Use real-time transcription as a license to stop taking notes. If the tool is capturing everything, you don’t need to write it down. That frees you to actually listen. That’s the highest-value use of real-time AI on a call.
Glance at the dashboard, don’t watch it. A quick look between questions is fine. Extended attention to the AI interface during the conversation is a problem.
Let competitor and objection flags come to you. If the tool surfaces something relevant, it’ll be there when you check. Don’t go looking for it mid-conversation.
The goal of real-time AI during a discovery call is to reduce the cognitive load of administration, note-taking, tracking time, and flagging topics so that more of your mental capacity goes toward the actual conversation. Used that way, it’s a genuine advantage. Used any other way, it becomes a distraction, wearing a productivity label.
After the Call: Where AI Earns Its Keep
If there’s one part of the discovery call process where AI delivers the clearest, most immediate value with the least downside, it’s the follow-up.
Here’s the standard rep follow-up process: call ends, rep has two more calls back to back, makes some mental notes, grabs lunch, sits down at 4 pm to write the follow-up email, can’t remember the exact wording the prospect used, writes something that’s close but not quite as sharp as the conversation was, sends it by end of day and hopes for the best.
Here’s what AI-assisted follow-up looks like: call ends, transcription tool generates a full summary within minutes, key topics discussed, pain points surfaced, next steps agreed, open questions remaining, and exact quotes from the prospect. Rep reviews it in three minutes, drops it into a follow-up email template, personalizes two sentences, and sends it within 30 minutes of hanging up.
The difference isn’t just speed. It’s accuracy and personalization. The follow-up email that uses the prospect’s exact language — the phrase they used to describe their problem, the outcome they said they were hoping for — lands differently than the one that paraphrases it. It signals that you were actually listening, ecause the transcript proves that you were.
The best follow-up email you can send is one that makes the prospect feel like you heard every word they said. AI-generated summaries make that easier to pull off consistently.
Beyond the immediate follow-up, call intelligence tools do something else that most reps underestimate: they create a searchable record of every discovery conversation you’ve had. Six months from now, when that prospect re-engages, you can pull up exactly what they said the last time you spoke. That kind of continuity is what eleven-year reps have by default. AI gives it to everyone else.
The Tools Worth Knowing
Gong. The enterprise standard for call intelligence. Transcription, real-time coaching, deal analytics, and manager visibility into call quality. Best for teams with a budget and a manager who will actually use the insights.
Chorus by ZoomInfo. Similar capability to Gong, often bundled with ZoomInfo subscriptions. Worth activating if your team is already paying for ZoomInfo and not using it.
Fireflies.ai. A more accessible price point than Gong or Chorus. Strong transcription, decent summaries, and good CRM integrations. A solid starting point for smaller teams and individual reps.
Otter.ai. The entry-level option. Excellent transcription, basic summaries, free tier available. If you’re not using any call intelligence tool right now, this is where to start today, for free.
ChatGPT or Claude for pre-call briefs. Don’t overlook the simplest option. A well-structured prompt, ‘Here is everything I know about this company and prospect. Generate a pre-call brief with key talking points, likely objections, and three discovery questions I should ask’, can produce a genuinely useful brief in under two minutes.
What AI Still Cannot Do on a Discovery Call
It would be easy to read this post and conclude that the discovery call is becoming automated. It isn’t. And it’s worth being clear about why.
The discovery call is not primarily an information-gathering exercise. It’s a trust-building exercise that happens to involve information gathering. The prospect is evaluating you as much as you’re evaluating them. They’re deciding whether you’re someone they want to work with, whether you understand their world, whether you’ll tell them the truth, and whether you’re in this for the relationship or the commission.
AI cannot build that trust. It cannot read the room when a prospect goes quiet after a difficult question. It cannot sense the shift in energy when you’ve hit on something that matters. It cannot make the judgment call about whether to push deeper on a painful topic or let it sit for the next conversation.
Those moments are yours. AI just clears the decks so you can show up for them fully.
What’s Coming in Post 4
You’ve found the right prospects. You’ve had a sharp discovery call. Now you need to write copy that actually gets responses and proposals that close.
Post 4 covers AI and sales copywriting: how to use AI to write in your buyer’s language without sounding like everyone else, the prompting frameworks that actually work, and the before-and-after difference between generic AI output and copy that converts.
Subscribe below if you haven’t already — Post 4 goes out next week, and it’s one of the most immediately actionable posts in the series.
One Thing to Do Before Post 4
If you’re not recording and transcribing your discovery calls, start this week. Otter.ai has a free tier. There is no reason to wait. Run it on your next three calls, read the transcripts afterward, and pay attention to the exact words your prospects use to describe their problems.
Those words are going to matter a lot in Post 4.
About This Series
The AI Sales Stack is a 7-part series on FillTheFunnel.com covering the tools, tactics, and workflows that are rebuilding the modern sales process. Miles Austin has been covering sales technology since 2006 and works with sales leaders and entrepreneurs to navigate the intersection of sales strategy and emerging technology.