AI-Powered Pipeline Management

In this IMPACT Summit 2025 research spotlight, I sit down with Walter Velasquez to explore one of the fastest-moving frontiers in GTM: AI-driven pipeline management. Walter opens with a provocation—“If AI didn’t capture it, it didn’t happen.” For decades, CEOs and CROs have relied on CRM data that depends on human inputs… which means human bias, human optimism, and human blind spots. AI changes the game by tapping the unbiased source of truth: what customers actually say, do, click, read, and use.

Walter breaks pipeline signals into three categories—product usage, engagement, and conversational—and shows how each aligns to different GTM motions (PLG, medium-velocity, and enterprise). Incumbents like Gong and Amplitude are adding “ask anything” layers and AI companions, while AI-first entrants are skipping straight to true agentic workflows that surface risks, qualify deals, and guide reps toward the right actions at the right moments.

Two guests make the research vivid. People.ai’s Natalie Wolf shows how AI-captured activity data eliminates “happy-ears forecasts,” collapsing 30-hour pipeline reviews into automated, scorecard-driven snapshots. Customers like Five9 saved over 1,000 hours per year, boosted qualification accuracy, and grounded forecasts in actual buyer behaviors—not anecdotes. Next, Pocus’s Sandy Mangat explains how AI agents monitor every account 24/7, cut the “toggle tax,” and give reps prioritized accounts, relevant signals, recommended contacts, and first-draft messaging. The results: 3.8× more closed-won and 2.8× more pipeline for reps using AI versus those who don’t.

We close with guidance for CEOs: start where the pain is, diagnose before you prescribe, and don’t let “open heart surgery” fears slow you down. Modern vendors layer on top of your CRM—not inside it—so time-to-impact is measured in weeks, not years. For companies serious about forecast accuracy, win rates, and efficient pipeline creation, AI-driven signal management is quickly becoming a top-priority investment.

Click here to access the written research.

00:00 — Let’s make pipeline management… fun?
Dave and Walter set the stage with humor and introduce the research.

01:29 — “If AI didn’t capture it, it didn’t happen.”
Walter’s opening salvo: CRM data is biased; AI listens to unbiased customer signals.

02:50 — Why human optimism breaks forecasts
Reps sincerely believe in deals; leaders can’t rely on hope-driven inputs.

04:02 — AI as copilot: surfacing what humans miss
AI flags missing personas, misaligned signals, or contradictions in the deal narrative.

06:41 — High-velocity sales and upper-funnel prediction
For fast cycles, signals before the deal hits pipeline matter most.

07:58 — Signals = customer truth
Dave summarizes AI reading emails, calls, behaviors—an unbiased observer.

08:26 — Defining “signal-driven pipeline management”
Walter introduces the concept: customers emit signals; AI translates them.

09:03 — Three types of signals (product, engagement, conversational)
Each aligns to a different GTM motion: PLG, fast-cycle sales, enterprise.

11:12 — Why this isn’t just Gong/Amplitude all over again
AI revolution: incumbents add “ask anything” and AI companions; AI-first players leapfrog with native-agent designs.

14:40 — Introducing the AI-first disruptors
Segues to People.ai and Pocus—post-2020 companies built for this moment.

16:10 — People.ai overview
AI that captures activity, maps it to pipeline, forecasts, and planning.

17:22 — The real bottleneck: manual pipeline management
Five9, AMD, and others spent hours logging data; leaders built forecasts by detective work.

18:33 — The transformation with AI
Automatic capture → structured scorecards → fact-based forecasting.

19:58 — 30 hours/week → still no confidence
Manual reviews were slow and unreliable.

20:55 — AI as the “unbiased source of truth”
Tracks personas, engagement, emails, and buyer intent signals.

21:11 — Building custom signals instantly
Health scores once took 6 months; now created in a day.

21:53 — Hard results
1,000+ seller hours saved; 100k+ activities captured; better coaching & CS alignment.

25:29 — Pocus overview
AI sales intelligence built around reps—prioritizing accounts using objective signals.

26:27 — Why start with the rep?
Reps are the frontline decision-makers; workflows break there.

27:19 — The toggle tax
Reps overwhelmed by siloed data across PLG usage, marketing, and calls.

29:01 — AI agents monitoring all signals 24/7
Agents pull intel from product, CRM, the web, transcripts—creating a unified view.

30:12 — Relevance ranking + recommended contacts
Only the most important signals rise to the top; messaging drafts included.

31:27 — Results
½ billion in pipeline; 50% of Asana’s pipeline; 3.8× more closed-won and 2.8× more pipeline for AI-using reps.

33:37 — “Do I really have to open my CRM chest cavity?”
Walter: no—AI tools layer on top; time-to-impact measured in weeks.

35:21 — Why pipeline management is high-impact, fast-impact
Unlike expansion/retention (annual cycles), pipeline responds quickly.

37:01 — AI + CRM layering explained
Best vendors plug into existing workflows; change management is modest.

37:51 — Should this be near the top of your AI list?
Yes—if pipeline is your biggest growth constraint.

38:55 — Diagnose before you prescribe
Always start with GTM motion, customer journey, and pain point.

39:29 — Fundamentals won’t change
GTM motions, customer journey mapping, framework-led diagnosis endure even as tools evolve.

40:22 — Wrap
Dave & Walter close the Spotlight; reminder: recurring revenue is architected, not earned.