Back to work

Instant Feedback

Instant Feedback — AI Impromptu Speaking Practice Software

For the frequent speaking-practice needs of arts-exam students, I packaged an AI practice product into a service that institutions could understand, purchase, and keep using, while maintaining its ongoing operation.

My contributionMy work extended from product packaging and institution communication to ongoing use and maintenance: helping students practise, parents understand progress, and institutions manage usage.

RoleProduct Operations Lead
TimelineMay 2026 to present
TopicsProduct packaging / Institution operations / User maintenance
~1 weekto first release
3institutions bought custom versions
~200students completed a practice
~100students kept using it

Background

Having a question bank does not mean students can keep practising.

Traditional arts-exam training relied on old question banks and manual feedback. Students could practise, but prompts changed slowly and practice rarely matched the real exam; teachers could not listen to and correct every response every day. The operational challenge was not only to explain what AI could do, but to help students start, parents understand the results, and institutions keep using the service.

Goals

Make the product usable and its value clear to institutions.

I translated the operational challenge into three directions: help students practise, help parents and institutions understand the value, and make the service purchasable and maintainable.

01

Help students start practising

Turn AI capability into a low-friction entry point

Make it possible to receive a prompt and start a complete practice at any time.

02

Make results clear to parents and institutions

Turn product value into understandable information

Let parents view progress reports and institutions manage student usage.

03

Make the service purchasable and maintainable

Package the product as a communicable, deliverable service

Maintain the product around institution-specific delivery.

Operations

I organised student practice, parent visibility, and institution management into one service.

I did not stop at explaining features. I organised the product around its users, institution needs, and delivery model into a service that could be understood, communicated, and maintained.

01

Product packaging

Structure the student practice flow

Turn prompts, preparation, recording, transcription, scoring, and progress records into a clear product path.

02

Three-sided service

Students practise · parents view · institutions manage

Organise the value of all three sides into a communicable, deliverable service.

03

First-release delivery

First release in about one week

Follow the product from idea to first release, focusing on completion and feedback in real use.

04

Institution operations

Communicate value and customisation

Own product explanation, institution communication, and customisation proposals; one month later, three institutions purchased custom versions.

Product demo

Students practise, parents view, institutions manage.

The student side shows six steps from prompt and preparation to recording, transcription, scoring, and progress records; the parent side shows a progress report; the institution side shows a student management view.

Interactive demo (all data shown here is fictional, for demonstration only)

An AI-powered arts-exam speaking practice and assessment system

I identify what AI can do,
then make it a practice system students can actually use.

Student side: six steps from questions to growth records.

Parent side: one-page growth report.

Institution side: one-page student management view.

No real accounts, recordings or scoring services are connected here.

Student · OriginalPrompt · Impromptu
即评大师9:41

Results & Review

From Product Adoption to Institutional Purchase

From the initial idea to the first release took about one week. One month later, three arts-exam institutions had purchased customised versions. Around 200 students completed at least one practice, and around 100 continued using the product.

Looking back, my role was not to explain the AI technology itself, but to turn the product into a service that could be understood, purchased, and maintained. I organised the value of the student, parent, and institution sides, followed the first release and real usage, and connected product explanation, institution communication, customised delivery, and ongoing maintenance. These results show that moving from “built” to “used and purchased” required more than feature completion: it required clear packaging, a communicable delivery model, and continuous operations.

01

First release

About 1 week

From idea to first release.

02

Institution purchase

3

Purchased customised versions one month later.

03

Completed practice

About 200

Completed at least one practice.

04

Continued use

About 100

Continued using the product.

My contribution

I turned an AI product into a service that could be sold and maintained.

My work covered product packaging, institution communication, user adoption, and ongoing project maintenance.

01

Product packaging

Owned

Turned AI capability into a service that students, parents, and institutions could understand.

02

Institution communication

Owned

Owned product explanation, institution communication, and customisation proposals.

03

User operations

Followed through

Tracked student completion and continued use.

04

Project maintenance

Followed through

Maintained product operation around real use, institution customisation, and follow-up adjustments.