# Chapter 2 — The Obvious Solution

## The Obvious Solution

🎧 [Listen to Chapter 2](https://thetwoproject.com/player.html?src=https%3A%2F%2Ftheowner-audio.s3.us-east-2.amazonaws.com%2FThe_Owner_Chapter_2.mp3\&chapter=2\&type=gitbook)

The pilot kicked off on a Thursday, in a conference room that smelled of dry-erase markers. Julia sat in the front row, arms crossed. The partners lined the back wall.

Lena Park, Northstar's AI and machine learning lead, stood at the screen, a laptop open and a clicker in her hand. She looked like three hours of sleep and pure momentum. Six months of unglamorous work sat behind this meeting: fighting the risk team, begging IT for GPU access, wrestling legal language into something that didn't sound like a confession.

"We start inside the firm," Lena said, projecting the first slide. It was simple: *Phase 1: Internal Acceleration*. "Drafting proposals. Drafting client emails. Drafting the first cut of analysis. We aren't touching client data yet. We're touching our own friction."

Tom Becker, leaning against the back wall, folded his arms tighter. His suit was impeccable. "Drafting our work product," he said.

"Drafting the first pass," Lena corrected, her voice sharp. "Humans still validate it. Humans still sign it. We are removing the blank page problem, not the thinking problem."

Tom's mouth twitched. "And when the first pass is wrong?"

"Then it's wrong the same way a first-year analyst is wrong," Lena said, clicking to a side-by-side comparison. "It gets corrected before it leaves the building. The difference is, the analyst takes two days to be wrong. The model takes two minutes."

A partner in the back—one Ethan recognized as the head of the Public Sector practice—spoke up. "Clients aren't paying us to be fast at writing, Lena. They're paying us to be right. If we send them hallucinated garbage faster, we just lose credibility faster."

Lena didn't look away from the room. Her eyes went straight to Julia. "Clients are leaving because we're slow to decide who does what. We spend a week formatting slides and three hours debating the recommendation. Apex isn't winning because they're smarter. They're winning because they move."

Tom's eyes sharpened at the mention of Apex. The wound from Orion was still fresh. "So your plan is to race them on slide production? We're a strategy firm, not a copy shop."

"My plan is to remove the excuse," Lena said. "If we can do the first draft in hours instead of days, we stop hiding behind 'effort.' We have to actually do the thinking."

Ethan watched from the side. Julia had asked him to sit in, to be the bridge between the technical reality and the partner panic. As a principal, he was senior enough to be blamed but not protected enough to survive it.

"Show me something real," Julia said, cutting through the murmurs.

Lena brought up a brief Ethan recognized immediately—Titanshield Insurance, last quarter. A deck they'd sweated over for three weeks, largely because Karen Holt didn't accept hand-waving and Compliance didn't accept anything without three citations.

"This is what we delivered," Lena said, tapping the finished deck. "Eighty pages. Three weeks. Four associates, one manager, one principal, one partner." She clicked. "This is a first pass generated from the meeting transcripts, raw notes, and the previous year's annual report."

The new deck appeared. It wasn't perfect. Ethan could see the seams—a few sentences that were too smooth, a transition that felt mechanical. But the structure was there. The argument was coherent. Even the appendix of policy quotes was populated, sorted by regulation code. In places, the model had gone beyond the source material — connecting a regulatory trend to a market shift that none of the meeting notes had made explicit. It was a leap, and it happened to be a good one. That was the power: the model didn't just retrieve. It synthesized.

Tom leaned forward, squinting. "How long?"

"Two hours," Lena said. "Including the time it took to scrub the confidential client data from the prompt."

"Two hours," Tom repeated, the number landing heavy in the room. "And you're telling me we still bill the same?"

Julia's gaze stayed on Lena. "What happens to quality?"

"Quality moves earlier," Lena said. "We spend human time on the decisions, not the formatting. We spend time verifying the logic, not aligning the text boxes."

Raj Mehta raised a hand from the second row. He looked like he hadn't slept since the Orion loss. "The junior teams love it. They stop doing midnight rewrites for a partner who's going to change the story at 8 a.m. anyway."

A few partners shifted uncomfortably.

Tom looked at Raj. "And the partner review? Do I still have to read eighty pages of machine text?"

Raj chose his words carefully. "It's… different. You have to review the thinking, not the grammar. It's harder, actually. The machine writes so confidently that you have to pay closer attention to spot the errors."

After the meeting, the partners filed out, muttering about billable hours and liability caps. Ethan walked with Lena toward the elevators.

"You're picking a fight," he said.

"I'm picking reality," Lena replied, closing her laptop. "Apex is sending clients next-day proposals. We take a week and call it rigor. Rigor is just a euphemism for bureaucracy."

Ethan watched the indicator lights over the elevator doors. "Speed is the easy part, Lena. Tom isn't wrong to be worried. If the machine writes it, who is responsible for the mistake?"

Lena's expression tightened. "Don't do this. Not yet. Let me get the pilot running before you start strangling it with philosophy."

"I'm not arguing," Ethan said. "I'm warning you. Every time we compress the work, we expose what we used to hide behind it."

"Good," Lena said, stepping into the elevator. "Let it show."

***

That evening, the compression showed up at his kitchen table.

Ethan walked into his apartment to the smell of roasted chicken and the sound of silence. Usually, the evening was a chaotic mix of cooking and homework panic. Today, it was calm.

His thirteen-year-old daughter, Maya, was sitting at the dining table, closing her laptop with a decisive snap. It was 6:15 p.m.

"Done?" Ethan asked, putting his bag down.

"Done," Maya said, sliding off her chair. "Mr. Henderson assigned a three-page essay on the Industrial Revolution. Causes and effects. It's handled."

Ethan looked at Sarah, his wife, who was cutting vegetables at the counter. Sarah was a history teacher herself, though at a different high school. She didn't look up. Her knife hit the cutting board in a steady rhythm.

"Three pages in twenty minutes?" Sarah asked.

"I did the research," Maya said, too quickly. "I found the sources. I just used the tool to help with the structure. It’s a tool, Mom. Like a calculator."

"A calculator doesn't do the math for you," Sarah said, finally looking up. "It does the arithmetic. This did the thinking."

"It didn't think!" Maya said. "I told it what to argue. I told it to focus on the urbanization aspect. It just... wrote it."

"She's skipping the struggle," Sarah said. "The essay isn't the point. The *writing* is the point. The writing is how you figure out what you think."

Maya rolled her eyes. "That's old-school, Mom."

She grabbed an apple from the bowl and went upstairs. Sarah cleared the plates without asking if anyone wanted seconds. Ethan sat at the table with his phone face-down, listening to the house settle into its separate rooms.

"It's coming for all of us," Sarah said from the kitchen. "If they don't have to write, they won't learn to think."

Ethan didn't answer. He was thinking about Tom Becker's question. *Do I still have to read eighty pages of machine text?*

***

Two weeks later, Ethan flew to Chicago to see Karen Holt.

Titanshield's headquarters overlooked the Chicago River. Karen met him in a corner conference room where the lake stretched flat and grey to the horizon.

Angela Ruiz, the Claims Operations Manager, sat beside her. Angela was the one who actually ran the engine room—she knew every workaround, every broken process, every compliant-but-useless procedure in the division.

"Julia says you want to talk about the new thing," Karen said, skipping the pleasantries. "And she says you're the one who can actually get it done."

Ethan took the opening. "We're testing an AI drafting tool internally. The results are significant. The obvious place to start with you is claims explanations."

Karen didn't nod. She sat perfectly still. "Start with the problem. Not the tool."

Ethan turned his screen toward them. "Your adjusters write denial explanations all day. It's high-friction work. Legal language, policy citations, empathy, and timing—all under load. If we can generate a first draft in minutes, we can give them time back for the hard cases."

"If it's wrong," Karen said, "we get sued. If it cites the wrong policy, we get fined."

"If it's right but inconsistent," Angela added, looking at her notebook, "we get escalations anyway. Right now, three adjusters will write three different denial letters for the same claim. That variance kills us."

Ethan let that sit. "That's why we don't auto-send anything. An adjuster reviews, edits, and approves before it goes out. This is about moving effort from typing to deciding."

Karen leaned back. "So you're asking me to approve a system that still requires my people to touch every customer message."

"I'm asking you to approve a pilot that reduces the time to produce a draft by eighty percent," Ethan said. "And then we measure whether review becomes the real bottleneck."

Angela's pen stopped moving. "We can measure that. We track review times now."

Karen's eyes stayed on Ethan. "Two questions. One: who is responsible for the outputs? Two: how fast can you start?"

Ethan answered carefully. "Operationally, this stays in your workflow. Your adjusters approve. Your supervisors audit. Michael Tran gets to set the rules for the prompts. Northstar builds and supports the pilot."

Karen didn't look satisfied, but she looked willing. "If Michael Tran doesn't sign off, it dies. And if an adjuster sends a hallucination, I want to know who approved it."

"Agreed," Ethan said.

"Two weeks," Karen said.

***

The next day, Ethan was in a very different conference room, in a very different part of the country.

Swiftcurrent Logistics headquarters was in a sprawling industrial park outside Atlanta. The air conditioning was fighting a losing battle against the humidity, and the noise of the nearby distribution center hummed through the walls.

Paul Jensen, the VP of Operations, didn't sit. He paced. Sitting down felt like downtime.

"Raj told me you have a magic button," Paul said, stopping to look at a dashboard on the wall. "He says it writes emails faster than my people can type."

"It's not magic, Paul," Ethan said, keeping his laptop closed. "It's a drafting tool. It uses your historical data to—"

"I don't care how it works," Paul interrupted. "I care about the bottleneck in Exception Handling. My guys are spending four hours a day explaining to customers why their shipment is late. I need them fixing the shipment, not writing apologies."

Ethan nodded. "We can automate the first draft of the update. Pull the tracking data, pull the reason code, generate the message."

"How fast?"

"Ideally? Seconds."

Paul stopped pacing. "And we can auto-send? If the data is right, just blast it?"

Ethan held up a hand. "I wouldn't recommend that. Not yet. The system can hallucinate. It can sound confident but be wrong about the location or the ETA."

Paul waved the concern away. "My people are wrong about the ETA half the time anyway. If it's ninety percent right and zero percent effort, I'll take the win. I have a backlog of two thousand exceptions. I need them gone."

Karen wanted control. Paul wanted velocity.

"We need a human in the loop," Ethan said. "Someone to approve the draft."

"Fine," Paul said, checking his watch. "But make it one click. I don't want them rewriting Shakespeare. Just 'looks good, send.' My Northeast regional manager, Rina Shah—she's already doing half the exception emails herself because her team can't keep up. Get her off that treadmill. Can we start Monday?"

"We need to set up the data connectors," Ethan said. "And we need to clear it with IT."

"I'll handle IT," Paul said. "Just give me the speed."

***

On the flight back to New York, Ethan looked out at the clouds.

Two pilots. Two weeks.

Titanshield wanted accuracy and was terrified of risk. Swiftcurrent wanted speed and was terrified of the backlog.

And back home, Maya wanted an 'A' without writing the essay.

Lena called it compression. Maya called it optimization. Paul called it throughput.

But Sarah's voice stayed with him. *The writing is how you figure out what you think.*

Karen Holt had asked the right question. *Who is responsible?*

Paul Jensen hadn't asked it at all.

Ethan opened his laptop and started drafting the pilot charter. He typed the heading: *Roles and Responsibilities.*

He left the section blank.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://thetwoproject.gitbook.io/the-owner-preview/ch02-the-obvious-solution.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
