The Agentic Age

The comment had a row number now.

Maren knew the words. She’d been carrying them for weeks: some cases look clean in the system but need a second read before the first message goes out. Eighteen of them, typed into a survey by a specialist she couldn’t name, read and reread until she could recite them. What she hadn’t seen, until this morning, was what the firm had done with them.

Desmond had found it, the raw adoption-pulse export, containing 612 responses logged during the project, then filed away when the license for the survey software expired. It had taken him two days to find it buried in the appendix of a closeout deck nobody had opened since it was presented. The spreadsheet showed the score of every response in the first five columns, with the comments in the sixth, and then, in the seventh and eighth, was the data Maren had come to see: each comment coded to a theme, and each theme assigned to a disposition.

He had it open on the second monitor in Maren's office. He scrolled to 487 and stopped.

The row read:

Confidence using the agent: 3 Ease of use: 5 Workflow clarity: 5 Would recommend: 5 Comment: Some cases look clean in the system but need a second read before the first message goes out. Theme: Exception guidance Disposition: Existing escalation pathway

Maren leaned closer.

"Well," she said.

Desmond waited.

"Someone read this." She was looking at the last two columns. "Coded it. Closed it. All before I got here."

"Months ago," Desmond said.

"Nobody ignored this." She sat back. "That's the part I can't get past. Somebody read it, understood it as a request for guidance, and routed it correctly."

She smiled slightly.

Desmond turned back to the screen.

Maren looked at the date.

Three days before project closeout.

“Would you show me what it became?”

Desmond opened the final adoption readout.

The document was twenty-eight pages long. He searched for exception guidance and found the phrase on page nineteen.

Specialists report high confidence and strong satisfaction with the new workflow. Isolated requests for additional exception guidance can be addressed through existing escalation pathways and future template optimization.

Maren read it.

Then she read the original comment again.

“That is a very good summary sentence.”

Desmond leaned back.

“Thank you.”

“You did not write it.”

“No.”

“Then I withdraw the compliment.”

He smiled.

Maren looked between the two screens.

The comment.

The summary.

“Both are accurate,” she said.

“Technically.”

“Yes.”

Desmond glanced at her.

“I thought we were respecting the numbers.”

“We are.”

“That sounds less fun every time you say it.”

Maren smiled.

“We are also going to respect the words.”

She pulled her chair closer.

“Do you know whose comment it is?”

“No.”

“You said last night you had a theory.”

“I do.”

“That is different.”

“I am learning from you.”

“I am so sorry.”

He laughed.

Then he pointed at the comment.

“I have heard Priya say almost the same thing.”

“So have I.”

“That does not make it hers.”

“No.”

“But you are going to ask.”

“Yes.”

Desmond looked at her.

“Kindly?”

Maren gave him a look.

“I know how to talk to people.”

“I have seen evidence both ways.”

“Thank you for the feedback.”

“You’re welcome.”

Maren opened a new spreadsheet.

At the top, she typed:

POST-LAUNCH SENSING

Desmond read it.

“That sounds cheerful.”

“It’s temporary.”

“I was worried.”

She made columns as she went, not the tidy kind she would put in a deliverable, just enough to hold the shape of the question. What the thing was. Who used to see it. Where it went when they did. When it stopped. What, if anything, stood in its place now.

"Those aren’t going to line up," he said.

"I know. I’m not trying to make them pretty. I’m trying to find out how many there were."

Desmond stopped smiling.

Maren looked at him.

“I know.”

“You think they all ended.”

“I think some of them did.”

“You know the manual review did.”

“I know it reduced.”

“You know the office hours ended.”

“I know the calendar series ended.”

“You know the pulse closed.”

“I know this was the final one.”

Desmond looked at the blank table.

“You are making me admire your restraint.”

“Please do not. I am very tired.”

He moved his chair closer.

“What is the first row?”

Maren looked at the raw comment.

“Second read.”

· · ·

They worked through the morning.

The first row was easy, because she had been living inside it for weeks. The manual second read. The specialists had done it on every standard case, once, before the first message went out, and it had caught the things no field recorded: the retirement that was a bereavement, the reduction the employee did’t know about yet. It had thinned after each expansion and mostly disappeared at full rollout, and what stood in its place was the agent’s own flags and whatever a specialist happened to escalate.

The second row took longer. The weekly case review, where the transition leads and operations and the implementation team had sat together and looked at what the week had produced. It had ended two weeks after full rollout, as scheduled, and what replaced it was managers raising things one at a time, through their own chains, to no one who saw them side by side.

The third row was the adoption pulse.

The fourth was office hours.

The fifth was the implementation steering group.

By late morning, the first four columns were mostly full.

The last one was not.

Maren looked at the gaps.

“Help me with the replacements.”

Desmond leaned toward the screen.

“I have been helping.”

“You have been sighing near the spreadsheet.”

“That is a form of support.”

“I need a more advanced service level.”

He looked at the first gap.

“What replaced office hours?”

“Managers.”

“How?”

“People asked their manager.”

“All right.”

Maren typed:

Individual manager response.

She looked at him.

“And if the same question showed up on three teams?”

Desmond was quiet.

Maren waited.

“No standing place to compare them,” he said.

She nodded.

“What replaced the weekly case review?”

“Escalations.”

“Where did they go?”

“Operations.”

“Were they reviewed together?”

“Not that I know of.”

“Across teams?”

“No.”

Maren typed:

Case-by-case escalation.

“What replaced the pulse?”

“The engagement survey.”

“How often?”

“Quarterly.”

“Agent questions?”

“No.”

She typed:

General engagement survey.

Then she sat back.

“I am not trying to make the closeout look negligent.”

Desmond looked at her.

“I know.”

“I mean it.”

“I know.”

“Projects end.”

“Yes.”

“Temporary structures are meant to end.”

“Yes.”

“And from everything we have seen, the closeout was competent.”

Desmond looked at the spreadsheet.

“That is the annoying part.”

“Yes.”

Maren smiled faintly.

“It would be easier if someone had done something obviously foolish.”

“I can find you some people.”

“Later.”

She looked at the last column for a while, the one that was supposed to say what replaced each thing, and then she started writing a note beside each entry, not a column at all, just something she needed to see on the page. The manager channel only worked if an employee or a specialist already knew enough to raise something. Escalation only worked if a specialist had already seen the problem. The engagement survey only worked if the employee tied what had happened to them back to the agent. The agent’s own flags only worked if the agent had already caught the case as an exception.

Every one of them started after someone already knew.

Maren stopped typing.

Desmond read the table.

Then he looked at her.

“Marcus triggered none of them.”

“No.”

“He did eventually.”

“On social media.”

“Not one of our approved channels.”

Maren laughed softly.

“That was almost funny.”

“I’m coming back.”

“I’m glad.”

Her phone buzzed.

Daniel.

I have the closeout materials. Twenty minutes?

Maren typed:

Please. And thank you.

Desmond read the message upside down.

“You added thank you.”

“I too am growing.”

“That’s concerning.”

· · ·

Daniel arrived with his laptop and a paper cup that had gone soft around the rim.

Maren stood.

“Thank you for finding all of this.”

“You’re very persuasive.”

“I have been a nuisance.”

“I was being polite.”

“I appreciated it.”

Daniel put his laptop on the table.

Then he noticed Desmond.

“Should I come back?”

“No, please stay,” Maren said. “Desmond is helping me reconstruct the handoff.”

Desmond said, “I thought I was sighing near a spreadsheet.”

“That too.”

Daniel sat.

Maren turned the table toward him.

“I know you closed this project months ago. I know the team had every reason to think the rollout had stabilized.”

Daniel nodded cautiously.

“And I am not trying to prove that someone forgot to do something obvious.”

“Good.”

“I am trying to understand what stayed open after the project team left.”

Daniel looked at the spreadsheet.

“Support?”

“Anything that could tell the firm whether the work itself was changing.”

“Usage reporting stayed.”

“I have it.”

“Manager reinforcement.”

“How did that work after the manager pulse ended?”

Daniel thought.

“Managers owned it.”

“How did they share what they were seeing?”

“They could raise issues through the functional leads.”

“And where did the functional leads compare patterns?”

Daniel looked at the table.

“We did that in the implementation meetings.”

“When did those end?”

“At closeout.”

Maren nodded and added the row.

Daniel watched her.

“What exactly are you trying to establish?”

“I am still trying not to decide too early.”

“That is unlike most consultants.”

“I have been humbled.”

Desmond said, “Repeatedly.”

“Thank you, Desmond.”

“You’re welcome.”

Maren looked back at Daniel.

“Can you walk me through business as usual?”

Daniel frowned.

“In what sense?”

“In practice. After the implementation team left, where did the different kinds of problems go?”

He thought about it.

“Technical issues to IT.”

Maren wrote.

“Process issues through operations.”

She wrote.

“People questions to managers and functional leads.”

Another row.

“Template changes through the content queue. Legal reviewed legal changes.”

Maren finished.

“Who looked across all five?”

Daniel looked at the list.

“The project team had.”

“And after?”

He was quiet.

“No one formally.”

Maren nodded.

“Thank you.”

Daniel looked surprised.

“For what?”

“For answering the question I actually asked.”

“I was not aware there was another option.”

“There usually is.”

She smiled.

Then she turned the spreadsheet toward him.

“This is what I am trying to understand. Not whether the closeout was wrong. Whether the project team was doing something no one realized would still be needed after it left.”

Daniel read the rows more carefully.

Maren gave him time.

Finally, he said, “That is not how we thought about it.”

“How did you think about it?”

“We were watching adoption.”

“Yes.”

“And adoption had stabilized.”

“How did you define that?”

“Usage. Confidence. Manager reinforcement. Office-hour attendance was falling. The issues coming in were minor.”

Maren pointed to the raw comment.

“Can I ask you something without taking it as a trap?”

Daniel looked at her.

“That is an alarming introduction.”

“Fair.”

She slid the comment toward him.

“Would you have coded this as exception guidance?”

He read it.

“Probably.”

“So would I.”

Daniel looked up.

Maren continued.

“You had six hundred comments. You needed categories. I am not going to pretend I would have run six hundred individual investigations because one sentence felt odd.”

Desmond looked at her.

“That is very honest.”

“I am trying a second new thing today.”

Daniel read the comment again.

Maren said, “What I cannot work out is what happened to the part that did not fit the category.”

Daniel looked at the final summary.

“She is saying some cases need a second read.”

“Yes.”

“That sounds like exception handling.”

“Maybe.”

Daniel looked at her.

Maren kept her voice gentle.

“An exception is something the process already knows is different.”

Daniel looked back at the comment.

“This says the case looks clean.”

“Yes.”

The room went quiet.

Daniel scrolled through the export.

“We could not have known what she meant.”

“I agree.”

He looked at her.

“Then what are you saying?”

“That I am starting to wonder whether knowing what she meant was the wrong requirement.”

Daniel frowned.

Maren pointed to the comment.

“She noticed something before she could explain it. Our process for receiving the feedback required the explanation.”

Daniel looked at Desmond.

Desmond said nothing.

Maren smiled faintly.

“He is being unusually disciplined. Please appreciate it.”

“I am.”

Desmond said, “This is humiliating.”

Daniel read the comment one more time.

Then he opened the closeout deck.

The final slide read:

TRANSITION TO BUSINESS AS USUAL

Below it were four boxes.

Technical Support: IT

Process Ownership: Operations

People Reinforcement: Functional Managers

Performance Reporting: Finance

Maren read the slide.

“Where did quality review sit?”

“Operations.”

“Can you show me?”

Daniel clicked backward through the deck.

Then forward.

“I need to find the operating model.”

“That’s fine.”

“No, I mean I thought it was here.”

“I know.”

He looked at her.

Maren said, “Please do not mistake my questions for accusations. I am genuinely trying to reconstruct this.”

Daniel nodded.

He kept searching.

Finally, he said, “I will have to check.”

“Thank you.”

He closed the deck.

“For what it is worth, nobody ended these things because they thought the work no longer mattered.”

“I know.”

“The implementation team was meant to leave.”

“I know.”

“The goal was a function that did not need a project team watching it forever.”

“I know.”

Daniel gave her a tired look.

“You say that a lot.”

“Only when I mean it.”

Maren looked at the spreadsheet.

“I am not trying to prove the team should have stayed forever.”

“Then what are you trying to prove?”

She thought about the wording.

“Nothing yet.”

Daniel waited.

Maren tapped the table.

“I am trying to find out what your team was seeing that business as usual had no mechanism to keep seeing.”

· · ·

Ruth arrived after lunch carrying the savings model.

Maren looked at the laptop.

“You came with the numbers.”

“I assumed you wanted them.”

“I always want them. I just rarely enjoy what happens next.”

Ruth sat.

“Good.”

Maren moved the sensing map aside.

“I need dates.”

“That’s refreshing.”

“The first realized capacity gain. The updated run rate. And when specialist intervention fell below target.”

Ruth opened the model.

“Which version?”

“The one Gerald took to the board.”

“Realized or projected?”

“Both, please.”

Ruth found the tab.

“The first realized capacity gain was booked in the second month after full rollout. The projected run rate was updated the month after.”

“What changed in the second month?”

“Volume increased. Average handling time fell. Manual touches fell.”

“Was manual-touch reduction part of the business case?”

“Yes.”

“Explicitly?”

Ruth looked at her.

“I do not hide assumptions.”

Maren smiled.

“I know. It is one of the things I like about working with you.”

Ruth paused.

Then looked back at the screen.

The compliment had nowhere obvious to go.

Maren let her keep it.

Ruth pointed to a row.

Specialist intervention rate.

Baseline: 61 percent.

Target: 20 percent.

Realized: 14 percent.

Maren looked at the number.

“Fourteen.”

“Yes.”

“And the capacity gain depends on that?”

“In part.”

“How much?”

Ruth looked at her.

“Do not do that.”

Maren stopped.

“You are right.”

Ruth’s expression changed slightly.

Maren continued.

“I was about to ask the question in the wrong order.”

“Yes.”

“Thank you.”

Ruth turned the screen toward her.

“The labor came out as manual intervention fell. That is true. The business case expected manual intervention to fall. Also true.”

“Yes.”

“You cannot say the reduction in intervention caused the Marcus Bell case.”

“I won’t.”

Ruth looked at her.

Maren added, “And I would like you to stop me if I get close.”

“I was planning to.”

“I assumed.”

Desmond, from the other side of the table, said, “This is a surprisingly functional relationship.”

Ruth ignored him.

Maren said, “Can we say the business case depended on specialists spending less time reviewing standard cases?”

“Yes.”

“And that the financial benefit appeared when that began happening?”

“Yes.”

Maren looked at the manual-touch trend.

“What did the quality model need to know about the cases no longer reviewed?”

Ruth considered the question.

Then she said, “Nothing.”

Maren looked up.

Ruth pointed to the model.

“It does not distinguish between a case that did not need review and a case that needed review but did not receive it.”

Maren nodded.

“That helps.”

“It shouldn’t.”

“It helps the analysis.”

“Fine.”

Maren wrote the sentence down.

Then she pulled the post-launch sensing table closer.

“What ongoing monitoring cost was included after closeout?”

“For adoption?”

“For human review. Case sampling. Cross-functional pattern review. Anything that continued testing whether standard cases were genuinely standard.”

Ruth searched.

“None.”

“Because it was out of scope?”

“Yes.”

Maren nodded.

“This is not a gotcha.”

“I know.”

“I want to be clear.”

“You usually do.”

Maren smiled.

“Was there capacity held for it?”

“No.”

“Funding?”

“No.”

“A trigger that would restore it?”

Ruth looked at her.

“What kind of trigger?”

“A complaint spike. A change in case mix. Rising overrides. An external incident. Something that would say: reopen review.”

“Not in the financial model.”

“Anywhere you know of?”

“No.”

Ruth looked at the sensing table.

Then back at the dates in her model.

“The model begins when the project cost ends.”

Maren stopped writing.

“Would you say that again?”

Ruth looked at her.

“The model begins when the project cost ends.”

She moved one finger along the timeline.

“Office hours. Temporary implementation labor. Additional review time. Project costs.”

“And after?”

“Realized savings.”

Maren sat back.

Desmond looked at the table.

Ruth closed the model.

Maren said, “Thank you.”

“For what?”

“For not making the answer more convenient.”

Ruth looked at her.

“Now that, I don’t know how to do.”

“I know.”

This time Ruth almost smiled.

· · ·

Maren found Priya late in the afternoon.

Chloe was sitting beside her.

The two women had the same case open on separate screens. Chloe was talking. Priya was reading.

Maren stopped several feet away.

Priya saw her and held up one finger.

Maren nodded.

“Take your time.”

Chloe pointed at the case.

“It is standard. I know it is standard. I just wanted to make sure I was not missing something.”

“What made you think you were?” Priya asked.

“Nothing.”

Priya looked at her.

Chloe tried again.

“The dates are strange.”

“How?”

“The role change was entered after the separation date.”

Priya read the file.

Then she opened a second record.

“That is because the role change was reversed.”

Chloe relaxed.

“So, it is fine?”

“The dates are fine.”

Priya kept reading.

Maren waited.

After another minute, Priya said, “Hold the first message.”

Chloe looked at her.

“Why?”

“The manager note says the employee does not know the role change was reversed.”

Chloe went back to the screen.

“Oh.”

“It is still a standard separation.”

“Okay.”

“But do not send the first message yet.”

Chloe nodded.

“What should I do?”

“Call the manager.”

Chloe wrote it down.

At the edge of the desk, she looked back.

“Would the agent have known that?”

Priya looked at the record.

“No.”

Chloe nodded once.

“Okay.”

She left.

Priya turned toward Maren.

“Sorry.”

“Please do not apologize. I was learning.”

Priya looked at her.

“From Chloe?”

“From both of you.”

Maren pulled over the empty chair.

“Do you have ten minutes?”

“Yes.”

“I can come back.”

“Maren.”

“Ten minutes.”

Maren smiled and sat.

She placed the printed survey row on the desk.

Priya read it.

Nothing in her face changed.

Maren waited.

“Did you write that?”

Priya read it again.

“I think so.”

“No rush.”

Priya looked at the date.

Then at the question.

“Yes.”

“You are sure?”

“I remember the survey.”

Maren nodded.

“All right.”

She sat back.

Priya looked at her.

“Is that all?”

“No.”

“I thought not.”

Maren smiled.

“I am trying not to make this feel like a deposition.”

“You’re doing better.”

“Thank you.”

Priya folded the page once, then opened it again.

Maren said, “Do you remember what you were trying to tell them?”

Priya looked at the comment.

“There was a question about confidence.”

“You gave that a three.”

“Yes.”

“Everything else was a five.”

“Yes.”

“Why?”

Priya looked embarrassed.

“I almost changed the three.”

“Why?”

“Because three looked dramatic.”

Maren smiled gently.

“On a five-point scale, it was practically a scene.”

Priya laughed.

Only once.

But she laughed.

“Everyone liked it,” Priya said.

“So did you.”

“Yes.”

“I recall you said the cases were moving. People were getting time back, and the messages were cleaner.”

Priya looked at her.

“You’ve been listening.”

“I have.”

The answer came simply.

Priya looked back at the page.

“I kept opening cases that were not flagged.”

“What made you open them?”

“I do not know.”

Maren waited.

Priya gave her a tired look.

“I have been trying to answer that question for months.”

“I can imagine.”

Maren softened her voice.

“And I’m sorry I keep asking you to produce an answer you did not have then.”

Priya looked at her.

Maren continued.

“Maybe try a different question.”

“All right.”

“When you wrote this, what were you hoping someone would help you with?”

Priya looked down.

For a while, she said nothing.

Then:

“I wanted someone to tell me how to know.”

Maren stayed quiet.

“When a clean case needed opening,” Priya said.

Maren looked at the comment.

Some cases look clean in the system but need a second read.

“That makes sense to me now,” Maren said.

Priya looked at her.

“I don’t think it could have made sense to the survey as written.”

Priya glanced at the coded theme.

Exception guidance.

“I was not asking when to escalate because escalation means you already saw something.”

“Right.”

Priya looked at Maren.

“I was asking about before that.”

Maren nodded.

Priya read the final summary.

Isolated requests for additional exception guidance can be addressed through existing escalation pathways and future template optimization.

“That is not what I said.”

“No.”

“What happened to it?”

“It went into the category available.”

“And then?”

“The pulse closed three days later.”

Priya’s eyes came back to her.

“The final one?”

“Yes.”

“And the office hours?”

“Five business days after closeout.”

Priya looked toward the floor.

People were working.

The queue screen showed green.

“I thought there would be another survey,” she said.

Maren did not answer immediately.

Then:

“I’m sorry.”

Priya looked at her.

“For what?”

“That nobody asked again.”

The words sat between them.

Priya looked away.

After a moment, she tapped the printed comment.

“This is not much.”

“It is eighteen words.”

“That’s what I mean. You cannot build a finding on one comment.”

“No, you can’t.”

Priya looked at her.

Maren smiled faintly.

“You can stop worrying. I have not abandoned all standards.”

“That is reassuring.”

“I am building it on everything around the comment.”

Priya looked down.

Then she asked, “Was I right? About what I was feeling?”

The question came quickly.

Maren looked at the woman who had just stopped a clean case because two dates did not fit.

“I don’t know enough yet to give you the answer I think you deserve.”

Priya’s face changed.

Maren continued.

“But you wrote this before Marcus.”

She put one finger on the date.

“Before anyone had a public case to explain. And that matters.”

Priya was quiet.

Maren left the raw survey page on the desk.

“You can keep that.”

“I already had it once.”

“Fair point.”

Maren stood.

At the edge of the desk, Priya said, “Maren.”

She turned.

“Thank you for asking differently.”

Maren smiled.

“Thank you for making me.”

· · ·

Daniel’s correction arrived before the end of the day.

Office hours ended five business days after closeout, not on the closeout date.

Maren replied:

Thank you. That distinction matters.

Tomas confirmed there had been no standing cross-functional quality review after the project team disbanded.

Ruth sent one sentence:

The business case explicitly depends on reduced specialist intervention in standard cases. It does not include post-launch funding for ongoing human sampling of those cases.

Maren read it.

Then Desmond appeared in the doorway carrying a printed memo.

“You are developing a habit.”

“What?”

“Standing in doorways with consequential paper.”

“I have range.”

He came in.

Maren looked at the memo.

“You found it.”

“I found three.”

“Of course you did.”

He handed her the first.

The title was:

Post-Expansion Review Concern: Declining Manual Intervention

The date was four months before Marcus Bell.

Maren read the first paragraph.

The reduction in specialist intervention is currently being treated as evidence of increased confidence in the agent-generated sequence. This interpretation may be premature. The same trend could also indicate that specialists are reducing review activity before the exception model is mature enough to consistently identify context-sensitive cases.

Maren read it again.

Then she looked at Desmond.

“You had this four months ago.”

“I know.”

“What happened?”

“You’ve seen what happened.”

“Show me anyway.”

He handed her the response.

Thank you, Desmond. Helpful observation. We will continue monitoring exception rates and escalation patterns as the rollout matures.

Maren read it.

Then she sat back.

“That is a good response.”

“Yes.”

“And That’s the problem.”

“Yes.”

They looked at each other.

Maren put the two pages side by side.

“I am sorry.”

Desmond frowned.

“For what?”

“For how long you had to carry this after receiving a perfectly reasonable answer.”

His expression changed.

Only slightly.

“That is a very specific apology.”

“I have been working on my range.”

He looked down at the memo.

“We monitored exception rates.”

“They went down.”

“Yes.”

“Escalations.”

“Down.”

“So, the concern appeared to clear.”

“Yes.”

Maren looked at him.

“What would have had to happen for it to stay open?”

“A failure.”

“One the existing system could see?”

Desmond looked at her.

“Yes.”

Maren opened Marcus Bell’s record.

No agent flag.

No specialist intervention.

No escalation.

No survey response.

Closed.

Desmond read it.

“The concern could only prove itself by getting past the thing that was supposed to stop it.”

Maren looked at him.

“That is close.”

“To what?”

“I am not sure yet.”

“Disappointing.”

“I thought you would appreciate the restraint.”

“I do not.”

Maren smiled.

She opened the memo beside the raw survey response.

Then the closeout deck.

Then the savings model.

Five documents.

Five dates.

Desmond watched her arrange them.

“You are going to use my memo.”

“Yes.”

“Do not lead with it.”

Maren looked up.

“Why?”

“Because then this becomes a story about someone ignoring me.”

“You were not ignored.”

“No?”

“You were answered.”

“Yes.”

“And the answer made sense.”

“Yes.”

Maren looked at the response again.

Desmond said, “If you lead with the memo, Gerald can fix the problem by promising the next Desmond a better meeting.”

Maren sat back.

“That is very inconvenient.”

“I know.”

“I had an excellent opening paragraph.”

“I’m sure.”

She smiled.

“Go on.”

Desmond pointed to Priya’s comment.

“I spoke.”

Then to his memo.

“She spoke.”

“Different channels.”

“Different shapes.”

Maren nodded.

“And nobody had a place to put them together.”

“No.”

“Not until Marcus.”

“Correct.”

Maren looked at the dates.

Desmond said, “Use those.”

“The dates.”

“Yes.”

“Nobody has to be bad for the dates to be true.”

Maren was quiet.

Then she said, “Stay.”

Desmond looked at her.

“I need you for this part.”

“That sounded alarmingly sincere.”

“It was.”

“I preferred the spreadsheets.”

“No, you didn’t.”

Maren pulled the spare chair closer.

“Help me make sure I am not forcing the evidence into the story I want.”

Desmond sat.

“That is a terrible job description.”

“It comes with coffee.”

“Your coffee is bad.”

“It comes with gratitude.”

“Worse.”

Maren smiled.

“Stay anyway.”

He did.

· · ·

They built the findings together.

Not the conclusion.

The sequence.

After project closeout, a standard case could receive additional review if:

The agent flagged it.

A specialist identified a reason to escalate.

The affected employee raised a concern.

A random sample selected it.

Maren read the list.

“Marcus triggered none of the first three.”

“And missed the fourth,” Desmond said.

She added it.

Before closeout, the firm had additional mechanisms that could surface a concern before it became an escalation:

Routine human second read.

Weekly case review.

Specialist office hours.

Adoption pulse.

Cross-functional implementation review.

Each ended during or shortly after rollout as planned.

Maren looked at the last two words.

“As planned.”

“Keep them,” Desmond said.

“I was going to.”

“You made a face.”

“I have learned everyone has opinions about my face.”

“I have several.”

“Noted.”

She kept the words.

Then:

Three days before closeout, a specialist wrote:

Some cases look clean in the system but need a second read before the first message goes out.

The comment was categorized as a request for exception guidance and closed against the existing escalation pathway.

Maren stopped.

Desmond read over her shoulder.

“You need her name.”

“The survey was anonymous.”

“You know it was Priya.”

“Priya told me.”

“Then ask.”

“I will.”

Desmond looked at the screen.

Maren added the benefit dependency.

Capacity gain.

Fewer specialist touches.

More cases proceeding without routine review.

Then the current controls.

Agent flag.

Specialist escalation.

Employee complaint.

Random sample.

She moved Desmond’s memo to the appendix.

He noticed.

“Good.”

“You were right.”

“I know.”

Maren looked at him.

“Enjoy that one. You’ve earned it.”

He smiled.

She returned to the first page.

The current evidence does not establish how many comparable cases have occurred.

That was Ruth’s sentence.

Maren continued.

It establishes that Marcus Bell’s case passed through all mechanisms currently in place without triggering any review.

Then:

Before any expansion decision, the firm must answer two questions:

What will identify a context-sensitive case before the first message is sent?

Who will review patterns that no single channel can see?

Maren read the page from the top.

Desmond did too.

After a moment, he said, “That is it.”

“Not yet.”

He looked at her.

“Please do not add six pages.”

“I was going to ask Priya.”

“Oh.”

Maren closed the document.

Desmond stood.

At the door, he stopped.

“Maren.”

She looked up.

“You did not need me to write that. You could have got there.”

“Eventually.”

He waited.

Maren said, “I would not have known what you kept.”

He looked at her.

“Or what you tried to say.”

Nothing.

“Or that Priya’s comment and your memo were the same concern arriving in different forms.”

Desmond looked away.

Maren continued.

“I could not have put this together without you.”

The wit left his face.

“Thank you.”

He nodded.

Then, because he was Desmond:

“Your coffee is still bad.”

Maren smiled.

“Go home.”

· · ·

Maren waited until after seven to call Priya.

Then she waited another five minutes because after seven felt rude.

At seven-ten, she called anyway.

Priya answered.

“Hello?”

“It is Maren.”

“I know.”

“Sorry to call this late.”

“It is not late.”

“Good. I have lost perspective.”

“What happened?”

“I finished the pre-read.”

Priya was quiet.

Maren looked at the survey page on her desk.

“I want to use your comment.”

“All right.”

“And your name.”

The silence changed.

Maren continued before Priya had to ask.

“You can say no.”

“I know.”

“I mean it.”

“I know.”

“I can use the comment anonymously.”

“Would that change the finding?”

“No.”

“Then why use my name?”

Maren looked at the document.

“Because the room needs to hear what you meant when you wrote it.”

“And for that they need me.”

“Yes.”

Priya said nothing.

Maren waited.

After a moment, she said, “I also need to be honest with you.”

“All right.”

“Your name may cause people to ask which cases you opened. Which ones you’ve changed, and whether you were working outside the documented flow.”

Priya was quiet.

Maren continued.

“I do not know what they will do with those questions.”

“Neither do I.”

Maren looked at the attendee list.

“I don’t want you to agree because you think I need you to.”

“You do need me to.”

“Yes.”

The answer came without hesitation.

“But not at any cost.”

Priya was quiet again.

Maren let the silence stay.

Finally, Priya said, “Use it.”

“Are you sure?”

“No.”

Maren smiled sadly.

“All right.”

“That was not a no.”

“I know.”

“Do not start.”

“Sorry.”

Priya laughed softly.

Then she said, “Use it anyway.” But Maren, don’t make me regret giving you permission because you regret asking.”

Maren closed her eyes.

“I will try not to.”

“That is not very reassuring.”

“It is the honest answer.”

“I know.”

The call ended.

· · ·

Maren reopened the pre-read.

She replaced:

A specialist wrote

with:

Priya Ramaswamy, Senior Transitions Specialist, wrote

Then she attached the raw survey page.

Desmond’s memo.

The closeout timeline.

The manual-touch trend.

Marcus Bell’s record.

She read the document once more.

No prevalence estimate.

No invented cost.

No claim that the agent did not work.

No claim that every clean case was bad.

No claim that anyone had knowingly removed a safeguard.

Only the dates.

The channels.

What each one could see.

And what none of them could see alone.

Maren sent the pre-read to Gerald, Ruth, Daniel, Tomas, and Desmond.

Then she created the meeting for the next morning.

Transition Operations: Findings Review

She added Priya.

For a moment, Maren’s cursor rested over Send.

Then she clicked.

The invitation went out.

Desmond accepted first.

Ruth second.

Daniel third.

Tomas fourth.

Gerald did not respond immediately.

Maren watched the screen longer than she wanted to admit.

Then Priya’s name changed.

Accepted.

Maren looked at it.

Priya Ramaswamy.

Accepted.

She closed the laptop.