Exam Workbench

The problem

Teachers already have large amounts of usable material in Word documents, PDFs, images, and scanned papers. The initial idea was to make exam formatting better, but market evidence showed that formatting, online question banks, drag-and-drop assembly, and printable layouts are already well served. The harder question is whether existing materials can become the starting point instead of a migration burden.

What existed before

The project began from a WPS exam-layout direction. That was technically useful, but the product thesis was too narrow: better formatting alone did not justify treating it as a major product. The work was reframed around a bring-your-own-materials workflow and explicit commercial kill gates.

The hypothesis

A useful wedge may exist if teachers can open material they already own, progressively select questions across sources, assemble a draft, and receive a high-fidelity editable DOCX plus printable PDF without first building or adopting a public question bank. Technical feasibility is necessary, but continued investment should depend on real-material, magic-moment, payment, and repeat-use evidence.

The first test

First prove the smallest end-to-end document path with real files: open two DOCX sources locally in the browser, select across them, reorder the draft, apply an A3 two-column page profile, export an editable DOCX, reopen it in WPS, and produce a printable PDF. Then test safe inline-image migration. Only after those technical risks are bounded should the project move into teacher-facing commercial validation.

What we built

A bounded browser-first prototype that can open two real DOCX files locally, select content from multiple sources, reorder a draft, apply an A3 two-column page template, export an editable DOCX, reopen it in WPS, and produce a one-page PDF. A second technical slice added safe migration of ordinary inline images between DOCX packages while explicitly failing closed on unsupported floating anchors, OLE objects, and layouts that do not fit the target column.

Evidence from use

The A0.1 vertical slice completed the full two-DOCX → cross-source selection → draft reorder → A3 two-column → editable DOCX → WPS reopen → PDF chain on real files. A0.2 then proved that ordinary internal inline images could be remapped across DOCX packages and rendered correctly in WPS. These are technical feasibility results only. They do not count as evidence that teachers prefer the workflow, will pay for it, or will return to use it repeatedly.

Where it stands

The browser form factor and a bounded multi-source DOCX workflow are technically plausible enough to justify commercial testing, but the product is deliberately still in Exploring status. The old thesis — that better exam formatting alone is the product — has been downgraded. The current investment decision remains gated on real teacher materials, a clear magic moment, real payment, and repeat use.

What we learned

• A technically solvable pain point is not automatically a strong product wedge, especially when major platforms already cover the visible feature.
• Commercial uncertainty and technical uncertainty should be tracked separately; a passing document prototype must not be rewritten as user demand.
• Starting from users' existing files may be more valuable than asking them to migrate into another question bank, but that remains a hypothesis until teachers choose it in real work.
• Failing closed on unsupported document structures is better than silently producing a paper that looks correct in the browser but breaks when reopened or printed.

Capabilities

Product validation · Document engineering · Browser-first applications · DOCX / WPS interoperability · Product strategy

The first idea was too easy to believe

The project started from a visible problem: teachers spend time formatting exam papers, and Word/WPS document layout is often tedious.

That problem is real. It is also not empty territory. Existing platforms already offer exam formatting, large question banks, drag-and-drop assembly, multiple paper sizes, and scanned-paper workflows.

So the product question changed from “can exam formatting be better?” to “is there a workflow that is materially better for teachers who already have their own materials?”

Prove the document path before building the product

The first technical gate used two real DOCX files rather than a synthetic editor demo. The browser opened both locally, content was selected across sources, the draft was reordered, an A3 two-column profile was applied, and the result was exported as an editable DOCX. That file was then reopened in WPS and used to produce a PDF.

The next slice tested a common source of false confidence in document tools: embedded images. Ordinary inline images were migrated between DOCX packages and accepted in WPS. Structures that were not yet safe — such as floating anchors and OLE objects — were left unsupported rather than being silently flattened or corrupted.

Technical PASS is not product PASS

The prototype answers one question: a bounded browser-first version of the workflow can work technically.

It does not answer the questions that determine whether this should become a serious business: will teachers bring real materials, recognize a meaningful advantage, pay, and come back?

Those remain explicit commercial gates. Until they are passed, this project stays Exploring by design.