Stop Using AI for Docs; Use It for Code Logic—Two Counterintuitive WorkBuddy Uses

I looked at Tencent's game job postings today, and honestly, quite interesting. The JD for "UE Micro-Horror Light Action Project - Senior Commercialization Operations" states: "Utilize creative design capabilities to design new character skins, peripheral resources, and other commercialization content for the game." Seeing these words, my first thought was—WorkBuddy can help you do half of this job, but you need to switch your usage method.

I know some people will criticize this. You're a game designer, teaching others how to use AI tools, specifically office tools—isn't that going off-track?

But seriously, after using WorkBuddy for three months, my biggest takeaway is: Most people lock themselves in by how they use AI tools. They let AI write documents and PPTs, then spend time editing and tweaking. Sure, it saves some time, but limitedly. The truly valuable usage is letting AI handle repetitive logical tasks you don't want to do, so you can focus your saved energy on things requiring your judgment.

For example, in a recent project, I needed to design a complex numerical distribution system. The numerical planner gave me a pile of requirement docs, saying, "Adjust the reward structure for this season, including but not limited to activity rewards, rank rewards, limited-time event rewards, and season pass rewards. The distribution rhythm, numerical curves, and resource types for each category need redesigning."

What does this sound like? Like a bunch of logic that needs organizing, calculating, and building into a rule system. Not like "writing beautiful copy."

I tried two approaches.

Approach 1: I wrote it myself. Opened Excel, dragged tables, wrote formulas, then copied and pasted into various module planning docs. Spent an afternoon, nearly went blind, and still missed calculating a resource production/consumption ratio correctly.

Approach 2: Threw it to WorkBuddy. But this time, I didn't ask it to write documents; I asked it to write code logic.

Here's the specific workflow:

Use Case 1: Turn Requirement Docs into Code Logic, Not Document Polishing

1. In WorkBuddy's chat box, paste the raw requirement doc I received. Note, don't ask it to summarize. Instead, tell it directly: "For this requirement, I need you to generate pseudocode describing the numerical distribution logic. Inputs are player activity level, rank, and current season time. Outputs are reward lists and their values."

2. WorkBuddy will first give a text description, which isn't valuable. What I want is the next step: In its Code Mode or Function Definition feature (Path: Bottom of chat box ... More Tools -> Code Mode), I say: "Help me write a function in Lua that calculates daily rewards based on three parameters: active_level, rank, and season_week. Reward types are daily_box, rank_bonus, and season_pass_point. Numerical values follow the distribution curve in Table 1 of the doc. For season_week, increase by 10% weekly for the first 8 weeks, then decay by 5% each time starting week 9."

3. WorkBuddy generates code. Then, don't just copy-paste. Take the code logic it generated and paste it directly into the "Implementation Plan" section of the design doc. The dev lead sees this and says, "Whoa, I can copy this logic directly."

Compare the effects:

Comparison Item	Writing Doc Myself	Using WorkBuddy for Code Logic
Time Spent	One afternoon (~4 hours)	20 minutes (including verification and adjustment)
Error Rate	Missed decay logic, numerical overflow	Complete logic, but curve parameters need manual fine-tuning
Dev Lead's Reaction	"I have to translate this doc into logic again, annoying"	"Great, I'll copy this directly"
My Gain	Wrote a bunch of text I didn't even want to read	Chatted with the dev lead for 30 mins, discussed a better numerical model

My Judgment: The correct usage of this feature isn't "letting AI write docs for you," but "letting AI translate the logic in docs into a language another team can use directly." For game designers, that means translating into program logic. For marketing, it means translating into executable marketing scripts.

Use Case 2: Make WorkBuddy a "Competitor Analysis Bot," Not Write an Analysis Report

Previously, my competitor analysis flow was: Find materials -> Read materials -> Screenshot -> Organize into PPT -> Write conclusions. In this flow, "Organizing into PPT" is the least valuable step, while "Reading materials" and "Writing conclusions" are the most valuable.

But most people let AI tools do the "Organizing into PPT" step, then gratefully start reading the PPT. That's wrong. You should let AI tools do the "Reading materials" step, so you can directly edit the conclusions.

Specific workflow:

1. In WorkBuddy, create a new "Knowledge Base" (Path: Left menu bar -> Knowledge Base -> New Knowledge Base). Dump all public materials, financial reports, player community discussions, and video guide links for competitors (e.g., Tencent's Delta Force overseas economic numerical analysis) into it.

2. Then, don't ask "Summarize this for me." Ask a precise question. For example: "Analyze this competitor's numerical distribution rhythm, focusing on the reward differences for 'non-paying players' and 'mid-paying players' in the first two weeks of the new season. Output Format: A table with two columns. Rows: 'Week 1 Daily Tasks', 'Week 1 Weekly Tasks', 'Weekend Events', 'Limited-Time Pass'. Comparison Dimensions: Reward type, numerical range, unlock conditions."

3. WorkBuddy retrieves relevant info from the knowledge base and generates a table. Take this table directly to align requirements with your project team, instead of spending two days digging for data yourself.

This is counter-intuitive: Good AI tools don't save you the "thinking" step; they save you the "information gathering" step, allowing you to double down your energy on "thinking" and "judgment."

A Counter-Example to Help You Avoid Pitfalls

One weekend, I had a brain fart and wanted WorkBuddy to "design an immersive experience event for a game." I gave it world settings and character backgrounds. It generated a bunch of very "AI-sounding" text, full of phrases like "In a corner of the map, a touching story about a 'Lost Civilization' triggers, players must make choices..." Put this in a game, and players will think, "Which intern wrote this?"

I closed the page and spent 30 minutes writing: "If your character is drunk in a tavern on a rainy day, an NPC nearby tells you his wife ran away with someone else. If you choose 'Buy him a drink,' he reveals the location of a hidden chest." AI can't do this. Because it requires "intuition for player experience," not "induction of text."

So, my current judgment is: WorkBuddy is best suited for "logic translation" and "information organization"—tasks with clear rules and boundaries—not "creative generation," which requires emotional resonance and intuitive judgment.

Finally, An Actionable Suggestion

Within this week, find the most annoying repetitive task you have. Examples: "Weekly sales data aggregation from three channels into a report," "Weekly competitor dynamic brief," or "Pulling user feedback weekly to organize requirements for product managers."

Don't let AI write the report directly. Instead, ask yourself: What is the most valuable "logical judgment" in this work? Is it "which data anomalies need flagging"? Or "which competitor dynamics are noteworthy"? Or "which user feedback should be prioritized"?

Then, hand over only the "data retrieval" and "initial organization" parts to WorkBuddy. Keep the "logical judgment" and "final conclusions" for yourself.

Trust me, use it this way for three months, and you'll find yourself leaving colleagues far behind. Not because you know how to use AI, but because you've invested time where it truly creates value.