The Room Where They Decide Challenge

What You'll Be Doing

The pilot was renewed. Meridian Freight is now deciding whether to roll the draft-quote assistant out to all 340 brokers across six service centers. You have one twenty-minute slot in the decision meeting.

In the room: Dana Whitfield, SVP Operations — non-technical, owns the budget, wants this. Priya Raghunathan, Director of IT — controls the infrastructure, and has said out loud: “You’re asking me to run your software in my account and page my team at 2am when it breaks.” And Meridian’s CFO, who has asked for one thing: payback stated in months.

Two architectures are on the table.

Option A — In-VPC deployment. Everything runs inside Meridian’s AWS account. Model calls leave through a private endpoint to your API. Nothing else leaves.

Build: 7 weeks
Run cost: $9,400/month infrastructure + $0.19 per draft quote
Requires 3 weeks of Priya’s team’s time during the 7. She has budgeted 1.
Passes Meridian’s security policy as written. No exception needed.

Option B — Vendor-hosted with a data gateway. You host the service. A lightweight gateway inside Meridian redacts and forwards.

Build: 3 weeks
Run cost: $2,100/month + $0.11 per draft quote
Requires a security exception: shipment records, including consignee name and address, leave Meridian’s network. Their written policy says no customer data leaves the VPC.
The exception is owned by Meridian’s CISO. Historically 9–14 weeks, and roughly 50% are approved on first submission.

The numbers you have been given.


Item	Value
Quote requests today	2,300/day (~50,000/month)
Projected after two shipper wins closing in Q1	3,800/day (~83,000/month)
Current broker time per quote	median 14 minutes
Broker time per quote with the assistant, measured in pilot	median 4 minutes
Share of drafts approved or edited-and-sent (not discarded) in the final pilot week	65%
Fully loaded broker cost	$61/hour
Dana’s budget line for this system	$28,000/month, all-in, fixed for year 1
Your firm’s rollout professional services fee	$180,000, one time

Make the call and defend it.

Constraints to consider

The $28,000/month is fixed for year 1. It does not move, and the CFO will not revisit it.
An option requiring a security exception cannot be presented as approved. If you recommend it, you state the probability and what happens if it is denied.
Whatever ships must be operable by Meridian’s own team after week 12. Your firm is not the on-call rotation.
You get one twenty-minute slot and you recommend one option. Not a decision framework.
You cannot change your firm’s per-draft pricing. $0.19 and $0.11 are what they are.

AI Usage Guidance

We expect you to use AI tools. We evaluate how you use them — not whether you use them. Evidence of iteration, redirection, and critical evaluation scores higher than a polished output with no process documentation.

The single highest-signal indicator: your video answer to the mandatory AI question. If you cannot name a specific moment where you redirected AI output, evaluators will assume you did not.

Mandatory AI question for your video: Walk me through one moment where you disagreed with, pushed back on, or redirected what the AI gave you — and what you did instead. Name the specific moment. Explain what the AI produced that didn’t meet the bar, what you did differently, and why.

Speak naturally. Communication is assessed on clarity of technical ideas and logical structure — not verbal polish, accent, or filler words.

What You'll Accomplish

Compare two deployment architectures on cost, time-to-value, risk and operability with real numbers

Reason about a security-review dependency as a schedule and probability problem, not a checkbox

State payback in a form a CFO can act on, and identify which variable actually decides the question

Build an operability position that survives an IT owner who did not choose the project

Defend an architecture recommendation to a non-technical buyer and a hostile technical stakeholder in the same room

How Your Work Will Be Scored

Architecture Choice & Trade-off Defense AI System Design in a Customer Environment Commercial & Stakeholder Judgment Video Walkthrough & Customer Communication AI Fluency

What to Submit

File 1 — Recommendation Brief

Document · 2 pages maximumRequired

Format: .pdf, .doc, .docx, .rtf, .txt, .md

Your recommendation, the trade-off argument with the numbers attached, and the rollout plan through week 12.

File 2 — README Document

DocumentRequired

Format: .pdf, .doc, .docx, .rtf, .txt, .md

Three required sections:

Section A — Written analysis: the economics. Run cost for each option at today’s volume and at projected volume, against the $28,000 cap. The payback calculation the CFO asked for, with your assumptions stated. And a sentence naming which variable actually decides this — because it may not be the one on the slide. 350–450 words.
Section B — Risk, operability and rollback: (i) how you handle the security-exception dependency if you recommend Option B, including what happens on denial; (ii) how the system becomes operable by Meridian’s team by week 12 — name the artifacts; (iii) your rollback position — what the trigger is, who calls it, and how long it takes.
Section C — AI Usage Log (Mandatory): This is not a trick. We want to see how you work with AI — not whether you used it. In a short section of your README, document your AI collaboration process. For each significant interaction with an AI tool, briefly note: what you asked the AI to help with / what it gave you / what you kept, changed, or rejected — and why. Three interactions documented is sufficient. The log does not need to be exhaustive.

File 3 — Video Walkthrough

Video · 6-8 minutetsRequired

Format: .mp4, .mov, .webm

Record as MP4 or MOV and upload directly on the Provn platform as a separate file. No ZIP files, no repository links, no third-party video platforms. Cover: (1) your recommendation and the one-line reason, ~60 sec; (2) the two architectures and the trade-off, explained the way you would say it to Dana, ~2 min; (3) the numbers — run cost against the cap, and payback, ~90 sec; (4) how you answer Priya’s objection, ~75 sec; (5) the mandatory AI question, ~60 sec; (6) the residual risk you are knowingly accepting, ~45 sec. Speak naturally, as if you were in the room — clarity of the technical argument matters, verbal polish does not.