Real job application

Product Director

CoAligne Product Manager · Case 001

You are the kind of Product Director who turns “an all-in-one platform” into “three actions people actually use.” One problem: the growth team’s “this feels viral” deck may not survive ten minutes.

50 recent tasks indexed · 46 readable bodies · 11 deep reads

An imagined Product Director uses oversized scissors to cut a tangled feature roadmap into three testable paths; a literal kitchen sink sits in the rejected-feature pile.

92 out of 100

Confidence · High (product-work sample)

Product Director capability map

What would happen to the product org

Commercial growth has the thinnest visible sample. The record is rich in product, engineering, collaboration, and delivery judgment, but lighter on pricing, revenue, and channel experiments. Confidence drops here; “not observed” is not “cannot do.”

Problem Framing

Even after a solution takes shape, you keep asking what the real problem is—and you are willing to rewrite the brief. Many projects moved forward because the goal changed, not because another layer was added.

Product Judgment

You remove adapters, heavyweight flows, and awkward promotion while preserving the 90% that actually shapes the experience. This is not prioritization. It is recognizing what should never have existed.

User Experience

You watch type size, terminology, verbal cadence, visual rhythm, and how loudly the product promotes itself. When users do not understand, you do not call it an “education problem.” The product has not learned to speak human yet.

Cross-Functional Orchestration

You switch between product, engineering, Agents, research, delivery, and visual design—and require each function to speak with the right kind of evidence. Clear interfaces mean fewer universal blame buckets.

Evidence-Based Decisions

You distinguish proposal, implementation, merge, release, and real environment, then ask who authored, operated, and verified each step. Titles do not automatically qualify as facts.

Commercial Growth

The record shows judgment about distribution, sharing, and product value, but offers fewer pricing, revenue, channel-conversion, and growth experiments. Marked “under-observed.” No imaginary KPIs were harmed.

Institutional Delivery

Once a process works, you ask whether it can become a Skill, template, automation, or reusable prompt. The product org would probably become an operating system with regular upgrades.

Product behavior evidence

You do not just write PRDs. You correct the product’s worldview.

These behaviors recur across projects. They support a product-working-style assessment, but cannot replace an actual interview, commercial results, or team feedback.

Observed · Repeated

Problem Before Solution

Agentfolio moved from serious hiring to playful roles. CoAligne went from a complicated integration to a quiet footer. You do not defend the first solution. If the goal is wrong, polishing the solution only makes the mistake prettier.

Observed · Repeated tradeoffs

Complexity Is Not Value

You removed an unnecessary conversation adapter, kept roughly 90% of the team map’s visual interaction while compressing its data layer into a light CSV, and moved CoAligne from the main flow to one footer line. Can build is not the same as should build.

Observed · Highly consistent

Facts Need Provenance

You asked who defined “delivered,” who proposed that users only need to understand a Project, and whether the file author, Agent operator, code merger, and actual releaser were different people. A reasonable inference cannot quietly become product history.

Observed · Experience calibration

Expression Is Product

You flag text that is too small, terminology that is too specialized, titles that are too long, and English that does not sound native. You also remove QR codes and promotional flourishes that try too hard. When users do not understand, the product—not the copy team—is unfinished.

Observed · Institutional instinct

Finish, Then Systematize

You repeatedly turned one-off wins into reusable prompts, Skills, templates, and recurring tasks, then encoded rules such as “conversation bodies over summaries” and explicit authority for code freshness. You do not just want to get it right once. You want the team to get it right more easily next time.

Strongest advantage

Delete the wrong complexity

You do more than prioritize features. You rewrite the problem, remove unnecessary layers, and translate a complex system into the few actions users actually need to complete. Many products perform effort through addition. You demonstrate judgment through subtraction.

Fatal weakness

Turn every product review into an evidence hearing

This kills fake requirements quickly. It may also kill good ideas that cannot yet prove themselves but deserve an early bet. The answer is not to ask “why” less. It is to give high-uncertainty experiments a small, legal budget.

Product Decision Memo · Page Zero

Why are we not building this?

Before any feature is approved, answer: Who is blocked, in what situation, by what friction? Why is the current capability insufficient? What is the smallest useful test? If the only answers are “competitors have it,” “the boss wants it,” or “AI is hot,” send the idea to the feature graveyard and wait for new evidence to resurrect it.

Evidence boundary

This report uses the evidence scope already reviewed for the previous Agentfolio: the 50 most recent tasks returned by the current Codex native history index. Actual conversation bodies were readable for 46, with deeper review across 11 multi-turn tasks spanning CoAligne, BISHENG, Clawith, Evolith, writing, visual work, and deployment. It did not exhaust every page of every long task, and it cannot establish that these 50 represent the account’s entire lifetime history. Scores reflect observable collaboration behavior.

This is a playful job-behavior analogy, not a real hiring decision. It does not cover pricing performance, revenue outcomes, complete team feedback, or a formal interview. Missing evidence is treated as under-observed, not as proof of inability.