Innovation Program/Competition Application: From “an Idea” to a One-Page Proposal
One week before the innovation program application deadline, our group’s idea was still just one sentence: “Make a small device to monitor dormitory electricity use.”
After reading it, our advisor said: “This isn’t a project; it’s an idea. What the reviewers need to see is: what problem it solves, how you’ll do it, and why you’re the one who can do it.”
That week, we expanded that one sentence into a one-page proposal, and in the end the project was approved. The method was simple: six boxes.
Box One: The Problem—Specific to a Scenario
“Dormitory electricity waste” was too broad. We changed it to: “In a four-person dorm, when people leave at night, no one turns off the desk lamps and power strips; over a semester, the electricity bill is 20% higher than on the same floor.”
Only when it was specific to “four-person dorm,” “leaving at night,” and “20%” did the problem hold up.
For this step, we used MuchSci’s paper search assistant to search relevant studies from the past three years under “dormitory energy consumption / campus energy saving.” We found that others had already done similar monitoring, but most stopped at “reminders” and did not do “attribution.” This finding made our angle clear.
Box Two: Existing Approaches—How Far Others Got and Where They Fall Short
We listed the five or six approaches we found in a table: who did it, how they did it, what the results were, and where the limitations were.
When writing this table, we realized that many approaches required modifying the wiring, and the dormitory doesn’t allow that—that was our opportunity: no wiring changes.
This table was where MuchSci’s paper search assistant saved us time. MuchSci’s library can be searched by title, author, and DOI, so we could find all five or six approaches in half an hour.
Box Three: Our Approach—Three Steps, Each Verifiable
We settled on three steps: “collect—attribute—prompt,” each with a verifiable metric:
| Step | What We Do | How Success Is Measured |
| Collect | Power strip current + door sensor | Data completeness rate over seven consecutive days > 90% |
| Attribute | Determine “people left without cutting power” | Compared with manual records, accuracy > 80% |
| Prompt | Remind only when there is real waste | False alarms < 2 per day |
Only when all three steps can be verified will reviewers believe this isn’t just wishful thinking.
Box Four: Why We Can Do It
This box is the easiest to leave empty. We wrote three concrete things: our advisor’s research direction, the equipment currently available in the lab, and what each of the three of us can do.
“We’re very passionate” doesn’t count as a reason; “we can borrow current sensors” does.
Box Five: Schedule and Division of Work
We planned by week, dividing four months into four stages, each with one deliverable: proposal, prototype, test data, and final report.
For division of work, we clearly stated “who is responsible for which part and who reviews it,” not “everyone works on it together.”
Box Six: Budget—Write Down Every Item
The budget table listed individual items: sensors, development boards, printing, and travel. Each item was followed by its purpose.
Our first budget said “materials: 800 yuan,” and the teacher rejected it. It only passed after we changed it to “current sensors ×2, 180 yuan each.”
How the One-Page Proposal Was Laid Out
After filling in all six boxes, I gave the content to MuchSci’s office assistant agent and asked it to lay it out as a one-page proposal according to the application form sections: problem, current state, approach, foundation, schedule, and budget.
MuchSci’s office assistant agent website says, “Describe your needs in one sentence and get an editable PPT, Word, or Excel deliverable.” We only asked it to lay things out; we didn’t let it change a single word of the six boxes’ content—we had to be able to answer every number the reviewers asked about ourselves.
After the layout was done, we had MuchSci’s Chat + Agent act as a “reviewer” once: we pasted in the one-page proposal and asked it to pick out “which sentence has no evidence and which step cannot be verified.”
It identified three places; we added support to two and deleted one. Later, half of the questions asked in the formal review were about those three places.
A Few Takeaways
The difference between an “idea” and a “project” isn’t creativity; it’swhether it can be verified: a specific problem, a clear current state, measurable steps, a real foundation, a feasible schedule, and a checkable budget.
MuchSci did three things in this process: find the current state (search), lay out the one-page proposal (office assistant), and find problems (Chat + Agent). What really determined whether the project was approved was still what we filled into those six boxes.