# Judgment, accelerated — executive presenter guide

## Bottom line

Show three moments where AI improves judgment: **before you commit, before you align a room, and while the team delivers**. Everything is pre-made except the clicks. The Lenny’s analysis engine gets extra time because it proves this is a purpose-built reasoning tool, not a clever prompt.

These are tested personal prototypes with invented data. Never imply that Chase, customer, employee, or production data was used. Production use belongs inside approved tools, sources, and controls.

## Five-minute preflight

- Open the enhanced URL with a fresh query string.
- Confirm the top bar says “enhanced edition.”
- Use 100% browser zoom and enter full screen.
- Press the right arrow through all six chapters.
- Test **Reveal revised decision**, **Trace one objection**, **Walk**, and **Open linked evidence** once.
- Return to `#start` and refresh.
- Open `roadmap.html` and `team-pulse.html` in backup tabs.
- Mute notifications and close unrelated tabs.
- If anything fails, speak to the artifact. Do not debug in front of the room.

## Recommended ten-minute path

| Time | Beat | Required action |
|---|---|---|
| 0:00–0:40 | Set the arc and guardrail | Click **Begin the story** |
| 0:40–2:20 | Ideate | Read the conviction and two questions; click **Reveal revised decision** |
| 2:20–4:20 | How the engine works | Explain build-once/query-time; click **Trace one objection** |
| 4:20–6:30 | Align | Click **Walk**, then **Send milestone request** |
| 6:30–8:40 | Deliver | Click **Open linked evidence** |
| 8:40–9:30 | Close | State human ownership and propose one bounded pilot |
| 9:30–10:00 | Buffer | Take one question or repeat the final sentence |

## Five-minute fallback

| Time | Beat | What to keep |
|---|---|---|
| 0:00–0:25 | Start | Arc plus invented-data guardrail |
| 0:25–1:25 | Ideate | Original conviction and revised conditional strategy |
| 1:25–2:20 | Engine | Archive → evidence → disagreement → citation |
| 2:20–3:20 | Align | Click **Walk** and name the four synchronized planes |
| 3:20–4:30 | Deliver | Open ownership evidence and state the human review gate |
| 4:30–5:00 | Close | Human ownership plus one bounded pilot |

## Exact talk track

### Start

> I use AI at three moments in product work: before I commit, before I align a room, and while the team delivers. It challenges assumptions, makes a strategy tangible, and finds patterns across work I could not read at once. I still own every decision. These are invented working examples—production use stays inside approved tools and controls.

Click **Begin the story**.

If interrupted: “The arc is Ideate, Align, Deliver. I’ll show the most distinctive part first.” Then advance.

### 01 · Ideate

> Most people ask AI to draft. I use it first to find the assumption I have not tested, then retrieve the strongest argument against me. Here, a broad distribution conviction was hiding two different causes of weak attach: discovery and workflow fit. The AI did not polish the slogan. It turned it into a conditional strategy with a six-week decision gate.

Click **Reveal revised decision**, then advance.

If interrupted: “The headline is that the output can now be proven wrong.” Move to the engine.

### 01A · The custom analysis engine

> This is the part I built, and the reason the result is different from a prompt. The system converted a frozen archive into more than fifteen thousand typed pieces of evidence, then connected more than four thousand pairs of genuinely opposing views. On a new draft, it brings in only the relevant defender, critic, condition, and source. It is a small reasoning system built around the work—not a model pretending to be four famous people.

Click **Trace one objection**, then advance.

If interrupted: “Extract once, retrieve the relevant tension, return a source trace.” Do not explain embeddings, batch APIs, or model names unless asked.

### 02 · Align

> The AI read the chosen strategy and translated every stage across four planes: the customer experience, the platform capabilities underneath it, the value unlocked, and the controls required. When I change stages, all four move together. People stop arguing about bullet points and start reacting to the same thing.

Click **Walk**, then **Send milestone request**, then advance.

If interrupted: click **Walk** and say, “Customer, platform, value, and controls evolve together.” Move on.

### 03 · Deliver

> This does not score people. It reads ticket prose, groups recurring system problems, links each theme back to evidence, and compares those patterns with an anonymous pulse. Here, the team ranked ownership fourth; the work record showed it was the second-most common delay pattern. The insight is not that the team was wrong. It may have normalized a chronic boundary.

Click **Open linked evidence**, then advance.

If interrupted: open evidence and say, “AI proposes the theme; a human checks every linked ticket before leadership sees it.” Move on.

### Close

> Across all three, AI handles interrogation and synthesis. I own the evidence, the decision, and the consequences. The next step is not a broad rollout. It is one bounded pilot in an approved environment, with read-only inputs, a named reviewer, and success and stop criteria agreed in advance.

Stop. Invite questions.

## Language discipline

| Avoid | Use |
|---|---|
| “I’m not waiting for permission.” | “I’m learning responsibly so I can help shape adoption.” |
| “What Chase tools can’t do yet.” | “An end state our approved toolchain can work toward.” |
| “AI does the work.” | “AI handles interrogation and synthesis; I own the decision.” |
| “Four real operators reviewed my memo.” | “The tool retrieved critiques grounded in their documented viewpoints.” |
| “Tickets show what actually happened.” | “The pulse captures experience; the record captures how work moved.” |
| “The team under-reported ownership.” | “The team may have normalized a major source of friction.” |

## Likely executive questions

**Is this already in production?**  
No. These are tested personal prototypes using invented data. They demonstrate the workflow. Production use requires approved tools, approved sources, security review, and the relevant control owner.

**Why not just use a general chatbot?**  
A general chatbot can draft. The custom engine prepares a domain-specific evidence system, retrieves opposing support, returns source traces, and has an explicit weak-evidence fallback. The quality comes from the context architecture.

**Are those experts really saying that?**  
The named critique is AI synthesis grounded in a dated documented viewpoint—not a direct quote or simulated participation. The source trail is visible, and a name is withheld if retrieval cannot support it.

**Could this work with internal knowledge?**  
Yes, if the source is approved and the use is appropriate. The same pattern can organize internal research, decisions, policies, or operating knowledge without moving it outside the approved environment.

**Does Team Pulse evaluate employees?**  
No. It is team- and system-level only. If it is used to score people, responses become less honest and the instrument stops being useful.

**What happens when the AI is wrong?**  
Important claims link to evidence and carry a status: verified, directional, or needs review. A human can reject or relabel them. Thin evidence produces an insufficient-data result, not confident filler.

**Why a clickable mock instead of a deck?**  
The room can see customer experience, platform work, value, and controls change together. It turns an abstract roadmap into one shared object people can test.

**What would you pilot first?**  
One bounded workflow with read-only inputs, a decision owner, a human reviewer, and explicit success and stop criteria. Napkin → Narrative is the lowest-data starting point.

## Hard stop

Never improvise a claim that real Chase data, customers, employees, or production systems were used. Never call AI synthesis a quote. Never offer to live-connect Jira or generate the full narrative in the meeting.