From Cold Call to Closed Deal: Using AI in Sales Conversations

Four words kill more deals than bad pricing, weak product-market fit, or tough competition: "Let me get back to you."

Every time a rep says it — fumbling for a pricing tier they can't remember, blanking on a case study that's perfect for this prospect's industry, or deferring a technical question they actually know the answer to — the deal loses momentum. Not dramatically. Not obviously. Just enough that the follow-up email gets opened two days later instead of two hours later. Just enough that the champion's internal enthusiasm cools. Just enough that a competitor's rep, who did have the number ready, sounds more credible.

This is a story about a sales team that decided to fix that problem — and what happened when they did.

The Information Problem in Sales

Sales calls share more DNA with job interviews than most salespeople would admit. Both involve real-time conversation under pressure. Both require recalling specific information on demand — pricing, metrics, case studies, competitive positioning. Both punish hesitation and reward confidence.

But here's where sales is arguably harder: the information surface area is enormous.

A typical B2B SaaS rep needs to hold in their head: pricing tiers across annual and monthly billing for three or four plans, discount authority limits by deal size, contract terms and SLA specifics, competitive positioning against five to ten alternatives, case studies organized by industry vertical and company size, integration details for dozens of platforms, product roadmap items they can and can't share, and the specific context of every open deal they're working.

Research shows that reps can typically recall accurate pricing for about 10 of their company's top 50 SKUs, and even those estimates fluctuate by 10% from the true price. The rest is approximation, or "let me confirm that and follow up."

The traditional fix is preparation. Print the pricing sheet. Review the battlecard. Skim the prospect's LinkedIn before the call. That helps — but it assumes you know exactly which information you'll need before the conversation starts. Sales calls are unpredictable. The prospect mentions a competitor you hadn't prepared for. They ask about an integration you haven't demoed in months. They shift from discovery to negotiation faster than expected.

This is the gap between static preparation and dynamic recall. And it's where 81% of sales teams are now experimenting with AI to close the difference.

The Pilot: A Mid-Market SaaS Team

In Q3 2025, a 12-person sales team at a mid-market SaaS company (B2B, $15K–$80K ACV) ran a quarter-long pilot with Neothi. The team was a mix of SDRs handling discovery calls and AEs running demos through close.

Before deploying, the team loaded Neothi with their full sales playbook: pricing sheets with discount matrices, the top 30 objection-handling responses from their enablement library, case studies tagged by industry and deal size, competitive battlecards for their five most-common alternatives, integration documentation, and each rep's open pipeline context from Salesforce.

They then used the overlay across three types of calls for 13 weeks. Here's what they found at each stage.

Discovery Calls: Better Questions, Fewer Missed Signals

The biggest change during discovery wasn't in the answers reps gave — it was in the questions they asked.

When a prospect mentioned a pain point, the overlay would surface relevant follow-up questions from the team's discovery playbook, filtered by the prospect's industry. Instead of defaulting to generic qualification questions, reps asked sharper, more contextual ones. A prospect in healthcare who mentioned compliance headaches would trigger suggestions about HIPAA-specific workflows. A prospect in fintech who mentioned scaling issues would surface questions about transaction throughput and audit trails.

The SDRs reported that they stopped worrying about remembering their question framework and started actually listening. One SDR put it bluntly: "I used to spend half the call thinking about what to ask next. Now I spend that half actually hearing what they're telling me."

This tracks with Gong's research on the optimal talk-to-listen ratio in sales calls. Top performers listen 57% of the time. But listening is hard when you're simultaneously trying to recall your next question, assess qualification criteria, and think about which demo to pitch. Offloading the recall part freed up cognitive bandwidth for the listening part.

Metric: Discovery-to-demo conversion rate went from 34% to 41%.

Product Demos: Eliminating "I'll Get Back to You"

Demos are where the information recall problem bites hardest. The rep is screen-sharing, walking through a product, narrating a story about value — and the prospect interrupts with a question about a feature they didn't plan to cover, an integration they haven't demoed in weeks, or a pricing question that doesn't map neatly to the standard tiers.

Every "let me get back to you" during a demo is a small crack in credibility. One is fine. Three in a 30-minute demo, and the prospect starts wondering whether this rep actually knows the product.

During the pilot, the overlay handled these moments by surfacing relevant information based on the conversation. When a prospect asked about Salesforce integration during a demo focused on the core product, the overlay pulled up the integration architecture and key data points — sync frequency, field mapping capabilities, known limitations. The rep didn't read from the overlay verbatim, but the nudge was enough to answer confidently instead of deferring.

The team tracked "let me get back to you" instances across all demo calls:

Metric: Deferred responses dropped by 60% — from an average of 2.8 per demo to 1.1.

The remaining deferrals were genuinely complex questions that required engineering input or legal review — the kind of "I'll get back to you" that's actually appropriate because it signals rigor, not ignorance.

Negotiation and Closing: Confidence With Numbers

Pricing conversations are where vagueness costs real money. A rep who hesitates on whether a 15% discount is within their authority — or who can't remember whether the three-year contract includes the onboarding fee or charges it separately — either gives away margin unnecessarily or kills the deal by suggesting they need to "check with their manager" on something that should be routine.

During the pilot, the overlay kept discount authority limits, contract term structures, and approval thresholds visible during pricing discussions. Reps could negotiate in real time without mentally flipping through their comp plan or opening a separate tab.

Top reps use data in 63% of successful objection responses. When a prospect pushed back on price, the overlay surfaced the relevant ROI case study or TCO comparison without the rep needing to search for it. The objection-handling wasn't scripted — the rep still had to read the room and choose the right response — but the supporting data was instantly available.

Metric: Average deal cycle shortened by 8 days. Self-reported rep confidence during negotiation increased from 3.2 to 4.1 on a 5-point scale.

What This Didn't Fix

Being honest about the limitations matters more than overselling the results.

Cold outreach performance didn't change. The overlay is useful during live conversations, not for cold email or LinkedIn prospecting. Outbound connect rates and response rates were flat during the pilot.

Reps who didn't prepare still underperformed. The overlay surfaces information you've loaded into it. Two reps who didn't upload their deal context or review their battlecards before the pilot got minimal value. AI recall requires material to recall.

Complex multi-stakeholder deals still stalled. When deals involved 4+ decision-makers, the bottleneck wasn't information recall during calls — it was organizational politics, budget approval cycles, and champion turnover between meetings. No call-time tool fixes that.

Onboarding took longer than expected. The team assumed reps would adopt immediately. In practice, it took about two weeks for reps to integrate the overlay into their natural conversational flow. The first few calls felt distracting. By week three, it was invisible.

The Compound Effect

The individual metrics — 7-point bump in discovery conversion, 60% fewer deferrals, 8 fewer days in the deal cycle — look modest in isolation. The compound effect is what matters.

More discovery calls converting to demos means more pipeline. Fewer deferred answers during demos means higher win rates from that pipeline. Shorter deal cycles mean the same team can work more deals per quarter.

Over the 13-week pilot, the team's total revenue was 23% higher than the previous quarter. Not all of that is attributable to a single tool — seasonality, a new product feature launch, and two strong enterprise deals played a role. But the team's VP of Sales attributed roughly half of the improvement to the combination of better discovery qualification and faster deal progression.

Bain & Company's 2025 research found that early AI deployments in sales boosted win rates by more than 30%, and sellers who effectively partner with AI tools are 3.7x more likely to meet quota. This team's results are consistent with those broader findings.

Who This Is For

The sales team in this pilot had three characteristics that made them a good fit for real-time AI assistance:

High information density. Their product had complex pricing, numerous integrations, and industry-specific use cases. If you're selling a simple product with one price point, the recall problem is smaller.

Frequent live calls. The team averaged 6-8 prospect calls per day. If your sales motion is primarily async (email, proposals), real-time call assistance is less relevant.

Existing enablement material. They had a well-developed playbook, battlecards, and case study library. The AI made that material accessible during calls — it didn't create it from scratch.

If your team matches that profile, the problem this solves is real and measurable. If it doesn't, there are probably higher-leverage improvements to focus on first.

Want to test it with your team? Start a free 14-day trial — load your playbook, run a week of calls, and compare.