Putting AI to Work — Review Deck

Talk track · 0:00

Melissa introduces Dan. No talking from Dan yet. Walk-on music plays. Dan settles, makes eye contact across the room before speaking.

Cue

• Wait for Melissa's intro to finish completely. Pause 2 seconds. Smile. Then start.

Talk track · 1:30 · iter 3 — show-of-hands open + ER anecdote

"Thank you, Melissa. [pause — eye contact, slow scan. Two full seconds of silence.]

Show of hands — how many of you are talking with AI every day? [pause — wait for hands. Don't fill silence.]

Yeah. [let it land] Almost all of us. Not typing. Talking. OK. [pause]

Jacco named the disease. Funnelism. I'm here to show you a cure. [pause]

Quick confession. Before I met Jacco — I was going to be a doctor. Worked in the emergency room. Loved it. [pause] Then I found go-to-market. [beat, half-smile] Turns out the ER was the calm one. [brief beat — don't sell the joke. Move.]

The next 30 minutes are about what every one of us wants to talk about — and most of us don't yet know how to. [pause]

One sentence from our 2026 briefing reframes the next five years. [gesture to slide]

'The next generation of revenue execution is not human-trained. It is human-designed and AI-delivered.' [pause — 3 seconds.]

What travels through a designed system isn't activity. It isn't pipeline. It isn't even revenue. It's signal — moving at speed. [slow on the three "isn't" beats — let the room feel the anaphora.]

Let me show you what that actually looks like."

Cues — iter 3 (Jacco SHIP WITH ONE FIX, applied)

• Jacco Show-of-hands open — terse. Walk on. Thank Melissa. Two full seconds of silence. Then ask. NO preamble. The Claude/GPT framing was padding — Jacco cut it on review: "Don't pre-frame what you're not going to ask. Just ask."
• "Not typing. Talking. OK." — three deadpan beats. Don't smile through this part. The room earns the moment by raising hands; you acknowledge it and move on.
• Jacco "I'm here to show you a cure" — INDEFINITE article. Humbler, inviting. Earns the rest of the deck — "what's the cure?" is now an open question the audience is listening for.
• Jacco ER anecdote — three beats, deliver flat, don't milk: setup ("going to be a doctor"), scene ("emergency room. Loved it."), punchline ("Turns out the ER was the calm one."). Brief beat, then move. Jacco on review: "The risk isn't self-indulgence, it's milking. Deliver flat, walk away."
• Through-line: disease → cure → ER doctor — Dan's lived experience grounds the medical metaphor. Audience will catch this subconsciously. Don't explain it.
• Jacco "Signal — moving at speed" — slow on the three "isn't" beats (activity / pipeline / revenue). Let the room feel the anaphora. Three-beat pattern, then "signal — moving at speed." Don't rush it.
• Jacco One time reference, not two. Kept "next 30 minutes" as session framing. Dropped the "next 25 minutes show you what running it actually looks like" closing — replaced with "Let me show you what that actually looks like." Jacco: "Pick one number. Two different timings in the same opener reads as you doing math on stage."
• ⚠️ Coordination note: S2 show-of-hands ("Talking with AI every day?") sets up S15 (Reasoning is Human) which currently asks the same question. When we refine S15, differentiate — S15 should pivot to "Who's TALKED with AI today?" or "Who's reasoning through a deal with AI?" Don't double-ask the same hands.
• DROPPED the Patterson/Kyle/Mark/Gabe/Radhika recap line — Jacco already paid the recap tax in his open. Show-of-hands replaces it.
• "Funnelism" — Jacco's sharpness. Single word, then move.
• Radhika callback still folded into S16 closing (loop closure: "Radhika told you it's hard. Julien just told you it's possible.").
• Time budget: ~1:25. Down from 1:32 after Jacco cut the Claude/GPT preamble. Cleaner than the iter-2 1:30 baseline.

Jacco-agent review — May 7 (iter 3 gut-check)

Verdict: SHIP WITH ONE FIX (applied: Claude/GPT preamble cut). Jacco's read: show-of-hands lands, ER anecdote earns its place barely (don't milk beat 3), a cure was the right edit, visual is doing the work, the 30/25-minute math reads as math on stage — pick one. All four caught and applied.

Talk track · 1:30

"Two years ago. 2024. The question was simpler. How do we get started. Pick a tool. Run a pilot. Show a productivity number. That was the bar.

The bar moved. Today the question isn't whether you started. Most of you started a year ago. [slow — plant the line] The question is — where did you put it, and does any of it talk to the rest of it? [gesture to slide]

INSEAD ran 515 companies through it. Same tools. Same access. The thing that separated the winners wasn't the tools. It was where they put them.

Most rooms put AI on the easy stuff. Email drafts. Meeting summaries. A bot here, an agent there. Real value — but stuck on the task. What actually reshapes how a firm performs requires seeing the system, not the task. [pause]

An AI application is one agent doing one job. A system is many of them, working together. That's what we're all trying to build right now — whether we know it or not. [pause]

So the doctrine isn't 'try harder' or 'buy more tools.' It's: see the system. Connect the system. Then deploy. Without that, AI doesn't make you faster. It makes you faster broken."

Cues — iter 3 (Jacco SHIP WITH ONE FIX, applied)

• Jacco Open cold: "Two years ago. 2024." NO "more useful question than 'are we behind'" preamble. Jacco on review: "You're announcing the question you're going to answer. Don't. Walk in and say it."
• Jacco "Most of you started a year ago" — say it slower. Plant it. The line that gives Dan permission to skip past the "are you behind?" anxiety.
• Jacco INSEAD line, de-jargoned: "INSEAD ran 515 companies through it. Same tools. Same access. The thing that separated the winners wasn't the tools. It was where they put them." Drops "controlled experiment" academic register. Saves 4 sec.
• Jacco "See the system. Connect the system. Then deploy." — anaphora triplet. The third beat breaks pattern (not "X the system") — that's the move. Don't touch it.
• Jacco "Faster broken." — two-word punch line. The line they'll repeat at lunch.
• DROPPED the "more useful question" preamble line — Jacco's iter-3 cut.
• DROPPED "local search" / "non-local" academic jargon — replaced with "Most rooms put AI on the easy stuff... What actually reshapes how a firm performs requires seeing the system, not the task."
• Tightened application/system bridge to 2 sentences (was 4) — sets up S4 cleanly.
• VISUAL: INSEAD pull-quote killed. Slide is now 18 words: headline + cyan period + anaphora subhead. Dan cites "515 companies" verbally — that's the proof, no on-slide attribution needed.
• Time budget: 1:30 — Jacco said keep it. S3 is the thesis; it earns its time.

Jacco-agent review — iter 3 gut-check

Verdict: SHIP WITH ONE FIX (applied: INSEAD pull-quote killed). Plus three voice-tuning catches all applied: kill the "more useful question" preamble; rephrase the INSEAD line without "controlled experiment"; slow down on "Most of you started a year ago." Triplet, "Faster broken," and 1:30 budget all stay.

Talk track · 1:30

"Picture what's happening in your company right now. [pause — slow]

Sales has built three agents. Customer success has built two. Marketing has one running outbound. RevOps has another scoring leads. Different teams. Different tools. Different formats — calls, emails, customer success usage data, transcripts, CRM fields. All of it, in different shapes. [pause]

Now ask the question nobody is asking out loud: how do those agents talk to each other? [pause]

Think about the world you already understand. When an app gets installed on your iPhone, the way it works at all is because there's an API and an iOS. Two layers that make sure your apps can connect.

Your AI right now is a pile of independent things. [pause]

It's going to get worse — because the speed is increasing. The agents are going to multiply. So whatever mess exists today, multiply it by ten. By a hundred. [pause]

The fix isn't more agents. The fix is a common language — one your agents speak, and one your humans speak. So we can read what the machines say. And the machines can read each other. [pause]

And it can't be a language the AI made up for itself. It has to be a language that we — the humans — designed, can test, can query, can correct.

What we've seen, when teams put that common language in place — close rate goes up. ACV goes up. Sales cycle goes down."

Cues — post-Jacco

• Jacco Replaces Table 4.1 entirely. Jacco was explicit on May 7: "you don't need it here." No revenue-architecture / growth-architecture talk.
• Jacco iPhone API+iOS analogy is Jacco's framing — "I had an API and an iOS that made it… made able that these two things connect."
• "Your AI right now is a pile of independent things." — say it slowly. The line that makes the room nod.
• "Multiply by ten. By a hundred." — sets up S7 PH502 later.
• DON'T say SPICED yet. The reveal is at S9. The temptation will be huge. Resist.
• The close rate / ACV / cycle drop is set-up for S5+S6, not the punchline.

Talk track · 0:25 · iter 3 — announcer cut

"[gesture to slide — let the room read it. Don't speak.]

'When everyone — and every agent — speaks the same language, we get better results.' [pause — 3 seconds]

Now you may ask yourself — what kind of better results?

Let me show you."

Cues — iter 3 (queued for Jacco batch)

• Jacco Don't announce the quote. Cut the prior "There's a sentence that captures what we just said" opener — silent gesture, let the room read, then read it aloud. Same anti-announcer rule Jacco applied on S3.
• Jacco Jacco's framing on May 7: "If we all speak the same language, guess what? We're gonna get better results. Now you may ask yourself, what are these results?"
• Read the quote once. Stop. Don't paraphrase.
• The "what kind of better results?" question is the bridge to S6.
• Visual: cyan period at end of quote. Quote text otherwise clean — no cyan keyword (would feel marketing-y on a community-proverb slide).
• Replaces the prior Ronen "Context compounds" pull-quote. Ronen still appears in S8 NOW tile.
• Time: 0:30 → 0:25 after announcer cut.

Talk track · 1:00

"We analyzed 52,000 sales opportunities across our customer base. Looked at every level of language adherence — sales and customer success. Three findings.

Reps who held the language all the way through had 2.1 times the close rate of baseline. Their ACV doubled. And their sales cycles got shorter. [pause]

Here's why it matters. The speed of selling and the speed of buying don't line up. Buyers go fast in moments — a strategic priority gets attention from a CFO, a quarter is closing, an event happens. They go slow when the buyer is ready to go fast. They go fast when the buyer is still going slow.

When your humans and your agents are all speaking the same language, the system can match the buyer's speed in the moments it matters. That's where the cycle comes out of the cycle. [pause]

52,000 deals. Same company population. The math is the math."

Cues — iter 3 (queued for Jacco batch)

• Jacco Buyer/seller speed mismatch — Jacco's note May 7: "the speed of selling and the speed of buying does not align well… you went slow when he was ready to go fast." Don't skip — it's what makes the sales-cycle drop legible.
• Read each number with weight.
• "The math is the math" — say it once.
• Iter-3 cuts: dropped "common" twice (S4 already established the context); cut "And in those moments, most sellers can't accelerate" (the anaphora pair carries the point); de-clinicalized "common-language adherence in the conversations" → "language adherence — sales and customer success".
• Visual: cyan period on "One pattern" — matches deck design language.
• Time: 1:00 → 0:57.

Talk track · 0:30

"And it only holds when you do it right. [gesture to slide — slow]

'AI doesn't fix broken systems. It multiplies them.' [pause]

If your agents don't share a language, AI doesn't reveal the dysfunction — it scales it."

Cues — iter 3 (queued for Jacco batch)

• Jacco Compressed earlier — Jacco's note: "you essentially made that point now." Don't relitigate.
• "Multiplies them" — sharp landing. Don't smile.
• Iter-3 cut: "There's a line in our internal manual that names what happens when you don't." — DROPPED (anti-announcer rule + dead reference: the manual citation was removed from the slide visual when 502→404).
• "If your agents don't share a language, AI doesn't reveal the dysfunction — it scales it." — KEPT per user direction. Dan's voice; lands on "scales it" before advancing to S8.
• Time: 0:30 → 0:22.

Talk track · 1:30

"Some companies are already there. They put the signal layer up. They got their agents and their humans speaking the same language. Three of them are in this room.

[Point to David Ronen tile — NOW] David Ronen, Anomalo — built the context layer first, pointed AI agents at it. Result: 146 percent lift in agent quality versus baseline. Not from a better model. From better context.

[Point to Charlie Wood — OVER TIME] Charlie Wood, Wiise — 450 percent recurring revenue growth across four fiscal years. NRR over 110.

[Point to Nida Ateeq — AT SCALE] Nida Ateeq, Lyra Technology Group — 91 acquired companies. One common language unified across all of them in under six months. [pause — sweep the room]

Three companies. Now. Over time. At scale. All in the room. [pause — sweep]

Now I want to ask you something. [advance to S9 — the question hero]"

Cues — iter 3 (queued for Jacco batch)

• Jacco "operating model" wording stays out — Jacco was emphatic May 7.
• Walk the three tiles in order: Ronen (NOW) → Wood (OVER TIME) → Ateeq (AT SCALE). Sweep the room after Ateeq. Then advance to S9 on "Now I want to ask you something."
• Iter-3 cleanup: trimmed S8 talk track to end at "[advance to S9]" — the question + turn-to-partner + bring-back + MEDDIC bridge all live on S9 only now (was duplicated across both slides after S9 was inserted).
• Iter-3 visual: removed redundant "Now · over time · at scale." subhead — the same labels (NOW / OVER TIME / AT SCALE) sit on each tile already; subhead duplicated them.
• Iter-3 visual: "91 OpCos" → "91 operating companies" (spelled out for non-PE audience).
• Iter-3 visual: cyan period on headline — matches deck design language.
• Time: 1:30 → 1:00. The 30 sec moves to S9 where the reflect-beat actually happens.

Talk track · 0:30 · REFLECT moment

(Dan advances to S9 mid-sentence — see S8 talk track. The question stays on screen while the room turns and talks.)

"What did you learn from your customers in April that's helping you win new prospects in May? [pause — 4 seconds, eye contact]

Turn to the person next to you. Take 30 seconds. Tell each other your honest answer. [wait — let the room actually do it. Resist the urge to fill silence.]

[After ~30 sec:] Bring it back. [pause]

Most of us can't really answer that. It's not that we're bad at our jobs. It's that the system that did April — customer success, renewals — doesn't speak to the system that's doing May.

AI-native companies ask a sharper version of that question. [pause] They ask: what did we learn yesterday? What happened this morning? [say it slowly — let the contrast land] Their signal layer is connected. [advance to S10 — MEDDIC + SPICED reveal]"

Cues — iter 3 (NEW SLIDE, queued for Jacco batch)

• Jacco The question gets its own page so the audience reflects on it while it's photographed. Photographable beat — phones will come up.
• Don't paraphrase the question on stage — read it as written, then stop.
• 30 seconds means 30 seconds. Look at your watch. The discomfort is the point.
• The slide stays up the entire time the room is talking. Don't advance until after "Bring it back."
• S8 talk track was trimmed iter-3 — the full question/turn/bring-back/MEDDIC bridge lives ONLY on S9 now (no duplication across slides).
• Jacco Slow on "what did we learn yesterday? what happened this morning?" — Jacco's line from May 7 review. Pause before. Pause after. Let the contrast (April/May → yesterday/this morning) land. This is the line.
• Iter-3 tightening: "Most of us..." paragraph cut from 4 sentences to 3. Same content, sharper parallel ("the system that did April... doesn't speak to the system that's doing May").
• Bridge to S10 (MEDDIC/SPICED reveal): "AI-native companies ask a sharper version… what did we learn yesterday? What happened this morning?"

Dan's bridge + reveal + question · 1:00

"So far I've called it a common language and a signal layer. [pause]

A lot of you in this room already use MEDDIC for sales — Metrics, Economic Buyer, Decision Criteria, the whole thing. MEDDIC is fantastic. For sales. But MEDDIC was built for sales — you can't onboard a CSM with MEDDIC. You can't use MEDDIC in marketing. You can't use it in customer service. It doesn't travel. [pause]

The signal layer has to travel. From the first outbound touch, all the way through onboarding, expansion, renewal, advocacy. And it has to be impact-centric — because what your customers actually want from you isn't a product, it's an outcome. [pause]

At Winning by Design we call that signal layer SPICED. Situation. Pain. Impact. Critical event. Decision. [pause — let it land]

SPICED travels. Left of the bowtie to the right. From outbound to advocacy. One language. [pause]

One company in this room ran exactly that play. They had MEDDIC in place. They added SPICED on top. And they let the signal travel. [Julien walks up. Dan shakes hand, gestures to S9 chrome.]

Please welcome Julien Cerutti, VP of Revenue Strategy at Meltwater. He rebuilt their enterprise GTM and CS end-to-end — MEDDIC plus SPICED — and he's now leading Meltwater's GTM AI initiative.

Julien — you took the 25 highest-risk enterprise accounts at Meltwater. The cohort that should have been mostly churned. Walk us through what you built, and what it produced."

[Dan steps stage-left. Julien advances to S10 to begin.]

Cues — iter 3 (queued for Jacco batch)

• Jacco THE SPICED reveal moment. Jacco's exact direction May 7: "SPICED is the only available signal that goes all the way from the left to all the way to the right."
• Jacco MEDDIC bridge: "MEDDIC is fantastic for sales but doesn't speak — you can't onboard a CSM with MEDDIC."
• "MEDDIC is fantastic. For sales." — the period in the middle is the joke AND the truth. Don't rush it.
• Tell Julien before stage: mention SPICED-as-connective-tissue ONCE in S11, ONCE in S13 close. Don't oversell.
• Ask the question once. Don't reframe it.
• Iter-3 fix: MEDIC → MEDDIC throughout (Metrics, Economic buyer, Decision criteria, Decision process, Identify pain, Champion). Was a typo in iter-2 baseline.
• Iter-3 cut: "Let me name it." — DROPPED (anti-announcer rule). Pause carries the reveal.
• Time: 1:00 → 0:57.

Julien · talk track · 1:30

"Thanks Dan. So — this started from the cohort no one wanted. 25 enterprise accounts, GRR-at-risk, mostly written off internally as 'going to churn anyway.' We pulled those out and asked one question: what's actually broken in how we're trying to keep them?

[gesture across the 5-step bar]

Five steps. Diagnose. Score and rebuild — 6-dimension rubric, baseline 55 out of 100, Joint Impact Plan 2.0 standard. Co-build — outcomes, KPIs, EB alignment with the customer. AI reinforcement — custom AI skill that scores the Joint Impact Plan and coaches the rep, always-on. Inspect and retain — embedded in CS rhythm.

Underneath all five: SPICED was the language that let MEDDIC, the Joint Impact Plan, and the AI all be talking about the same customer. The signal could travel.

The result: +7.5 points of GRR. Joint Impact Plan quality 55 → 67 (+22%). Flagship account 83 / 100. [advance to S11]

Now let me show you what we found in the audit — and what Joint Impact Plan 2.0 fixed."

Cues — post-Jacco

• Jacco SPICED-as-connective-tissue line is a Jacco ask May 7. ONE sentence is enough. Don't lecture.
• Land hard on +7.5 pts / 55→67 / 83/100.

Julien · talk track · 1:30

"On the left — what we found in the audit. Vague outcomes — Joint Impact Plans described deliverables, not future-state business impact. No measurable KPIs. Weak stakeholder depth — economic buyers missing, single-thread risk high. No cadence or accountability. Superficial co-build — Joint Impact Plans authored by Meltwater, not built with the customer.

[gesture across to the right column]

Joint Impact Plan 2.0. One slide per outcome, specific, measurable, linked to an executive priority. Before, after, impact. Provable at renewal. Economic buyer named and multi-threaded. Dated milestones with owners. Customer fingerprint throughout. Co-build is non-negotiable.

[advance to S12]

But the standard alone doesn't scale. Here's the AI piece."

Cues

• Slide labels are minimal — your voice carries the descriptors.
• Visual contrast (red vs. cyan) does the structural work.

Julien · talk track · 1:30

"The best system is one people use without being told to. So we built a custom AI skill that scores every Joint Impact Plan against that 6-dimension rubric. Reps submit plan plus a video walkthrough.

[gesture to scorecard]

Six dimensions, weighted. Outcomes 25%. Metrics 15. Stakeholders 20. Strategic alignment 10. Co-build 10. Next steps 15. Flagship account scored 83 of 100. Top quartile.

[gesture to right-side boxes]

What the rep gets back is three things. Assessment — diagnosis-led, the single most important gap, framed like a coach. Score — 6 dimensions, weighted, dimension-level reasoning. Next step — one execution thesis, actionable in 24 hours. [pause]

The bigger insight isn't the +7.5 points. It's that we took the same operating rhythm — SPICED-grounded outcomes, MEDDIC where it fits in sales, AI reinforcing the standard, humans inspecting at cadence — and ran it on the customers we'd given up on. They outperformed our average. That's what a signal layer does. And that's why we're doubling down on it."

Closing transition (Dan)

"Thank you, Julien." [Pause for applause. Julien exits. Then advance to S13.]

Talk track · 2:30 · Jacco's narration verbatim where powerful · iter 3 — Jacco SHIP WITH ONE FIX applied

"Julien just showed you what happens when the signal travels. Now I want to talk about how we're willing to be paid for it. [pause]

For a decade, most of us in this room sold or bought subscription software. And what we learned — what our customers learned — is that a lot of users didn't get the value they were promised. Or, more honestly: they didn't get the impact they were promised. [pause]

That's why your customers want impact today. Not a tool. Not a license. Not a seat. Impact. [pause — slower]

We at Winning by Design experienced this ourselves. Historically, we trained. And we've come to realize: the value of training was mostly perceived as 'I got a certificate.' Not 'I delivered.' The whole nature of training — across this industry, for a hundred years — hardly ever connected directly to deliverability. [pause]

So we realized: in this new economy, the impact-centric economy, we too must change. That's why I'm telling you what we're announcing. [pause — slide reveal]

We are changing our pricing model. We now offer a Pay Per Impact option. [pause — slide reveal lands]

Here's how it works. The infrastructure of our customers has changed. Calls can be recorded. Most of yours are. Ten years ago we couldn't count on that. After COVID, we can. Gong, Fathom, Zoom — universal. That gives us a new ability. We can recognize, from the call recordings themselves, whether the word 'impact' was actually used.

Because the new economy is recurring revenue as a result of recurring impact. [pause]

And here's what we've learned, in our data and our customers'. When sales and customer success organizations actually talk about impact in their conversations — not the product, the impact — they get 59 percent more revenue out of the same customer group. Same customers. Same product. Just by changing what gets talked about. [pause]

This is a stark contrast from a decade ago. A decade ago, growth got squeezed out of the decision stage. Decision criteria. Decision process. Decision maker. Decision, decision, decision. But how do you pay a salesperson for closing a decision when the world has moved to paying for impact? [pause]

It's no longer about the decision. It's about delivering impact. We need to train differently. Price differently. Operate differently. [pause]

That's why we're introducing Pay Per Impact pricing. And we encourage you to think the same way about your business — because if you don't, your customers are going to find someone else who does."

Cues — post-Jacco

• Jacco Entire talk track is reverse-engineered from Jacco's May 7 narration. Where his phrasing is sharp, it has been preserved verbatim.
• "I got a certificate, not I delivered" — Jacco's. Say it the way he would.
• "Decision, decision, decision" — three quick beats. Then pivot.
• "59 percent more revenue out of the same customer group" — read with weight. Only proof point on this slide.
• Closing line — "if you don't, your customers are going to find someone else who does" — replaces "we trust the system because we engineered the variance out." Tighter.
• Pacing: this slide above all others wants Jacco's "talking to a person from five years ago" register. Slow.

Talk track · 2:30 · Kai video GO (May 7)

"Quick question. Who in this room has TALKED with AI today already? [pause — wait for hands. Different from S2's "every day" — this one's same-day, sharper.]

Yeah. Almost all of us. In the car. Walking the dog. On the treadmill. In between two meetings. Sitting on the couch at 10pm with a question that's been bugging you all day. [pause]

And what is that, when we're doing it? It's reasoning. That's a human skill. We learn through conversation. We always have. [pause]

So why aren't we letting our salespeople do that with their deals? Why are we still asking them to wait for a Monday morning one-on-one? Their best reasoning happens at random moments — five minutes before the call, twenty minutes after, the night before, the drive home. The agent has to be available when the rep wants to think, not when the calendar allows. [pause]

1mind built that agent. Sales Coach Kai is the agent that coaches the signal until it lands. Available 24/7. No judgment. Just a peer to reason with — about a real deal, in real time.

Watch this for sixty seconds." [Play Kai video — 60–75 sec.]

"1mind and Kai are both in the lobby — right outside this room — for the next two hours. Go talk to them about your deals."

Cues — post-Jacco

• Jacco "Reasoning is a human skill. We learn through conversation." — Jacco's exact framing May 7. Don't paraphrase.
• Jacco Raise-of-hands moment is real audience engagement. Wait for hands.
• Don't recite a partner roll-call. 1mind = carrier, Kai = coach.
• Kai video tighter and faster than v1 — see KaiScript.md update May 7. Jacco's note: "don't ask the longest fucking questions."
• If video glitches: cut at 20 sec — "Kai is in the lobby. Go meet him."

Fallback (no video)

"1mind built the agent that carries the signal at scale. Sales Coach Kai coaches the signal until it lands. Both are in the lobby right after this session. The conversation in the lobby will go further than any 60-second clip I could play here."

Talk track · 2:00 · REPLACES "three things"

"I want to ask you something before we go to lunch. [pause — slower than feels natural]

Do you remember the moment in early 2020 when COVID became real for you?

Not when you read about it in the news. Not when the case count was rising. The moment. When it crossed from 'this is happening over there' to 'this is happening to us.' [pause — eye contact]

For a lot of us it was a single image. The empty highway. The first time you saw the National Guard staged at a hospital. The afternoon you went to the hardware store with your kids and bought wood — wood — to board your windows in case things got worse. The dinner where you sat with your family and somebody said, out loud, 'I don't know how long we have.' [pause — 3 seconds]

We were ignorant up until that exact moment. It was all announced. The case curves were public. The travel restrictions were already on the news. And we still didn't really get it — until the day it knocked on the front door. [pause]

Folks. [pause — long]

It is May 12th, 2026. [pause]

AI is knocking on your door. It is happening right now. We are months — not years — away from the equivalent of the empty highways. The moment when whatever you didn't get around to building, you don't get to build anymore. [pause]

If you haven't put up the signal layer — if your agents and your humans aren't speaking the same language about the same customers about the same impact — and you don't start this quarter, you are going to be late. Not behind. Late. [pause]

So before you eat lunch, do three things — quick, and then we go.

One. Sales Coach Kai and the 1mind team are in the lobby, right outside this room. Go meet them. Not after lunch. Before lunch.

Two. Several of the people you saw earlier — Ronen, Wood, Ateeq — and other operators in this room are nominated for Growth Awards this afternoon at 3:50. Every nominee in this room built the signal architecture before they multiplied it with AI.

Three. Find the person sitting next to you at lunch. The one you turned to a few minutes ago. Tell them what you're going to build first when you get back. Out loud. Make them remember you said it. [pause — slow]

Radhika told you it's hard. Julien just told you it's possible. [pause]

The ones who figured it out — they know what their customers learned in April. They know what their customers learned yesterday. They know what happened this morning. Their agents speak. At velocity. That's what the signal layer makes possible. They escaped Funnelism on purpose.

If Pay Per Impact resonates and you want to talk — I'm Dan Smith. dan@winningbydesign.com. Or stop me at the Growth Bar. [pause — eye contact]

It is May 12th, 2026. AI is knocking. Build the infrastructure now.

Thank you."

Cues — post-Jacco

• Jacco ENTIRELY new close — Jacco was unambiguous May 7: "you don't send them to lunch with three things, dude. You want them, like, you want them to go into lunch, go like, holy crap."
• Jacco "Activate their visual brain" — empty highway, National Guard, wood for windows, family dinner. Memorize, don't read. Live these images.
• "It is May 12th, 2026" — say the FULL date slowly. The audience needs the parallel to the January 2020 anchor.
• The three lunch items still happen — but framed as "go light a fire and come back," not as a tidy checklist.
• "Radhika told you it's hard. Julien just told you it's possible." — closes the open loop from the cold open.
• "Thank you" — said once, with finality. Don't add anything after.
• DO NOT smile during the COVID memory. Smile after "Thank you," not before.

Talk track · —

Lights up. Melissa returns to stage. Dan bows briefly, exits stage right.