Make the product easier to define.
Ghostpen could take an idea, recording, document or link and generate different formats. The harder question was what it should do exceptionally well.
I focused the direction on one workflow: use the source, decide what to make, review the work and return to it later.
- 01SourceBring existing knowledge
- 02DirectionChoose audience and purpose
- 03WorkCreate the useful pieces
- 04ReviewEdit and approve
- 05ReuseReturn without starting over
Give each part a clear job.
Start with what people have.
A recording or document already holds useful context. I made that material central to the workflow instead of asking people to rebuild it in a blank prompt.
Make the right few things.
I pushed the roadmap towards intentional packages. More output formats were not useful if people could not finish, review or reuse the work.
Keep approval explicit.
People needed to correct the source interpretation, edit generated content and decide what to approve. AI suggestions could not quietly become final decisions.
The work should survive the session.
I organised the product around sources, campaigns, artifacts and decisions. That gave web, mobile and backend a shared model for saving and continuing work.
Mobile focused on capture, review, approval and returning to saved work. Deeper production could stay on a larger screen.
Check the whole journey.
I worked across design, web, mobile, backend and QA. We checked authentication, uploads, processing, editing, saved work, approvals and error recovery.
This validates the tested technical paths. It does not establish market fit.
Let user behaviour guide the roadmap.
The product direction and shared model became clearer. The commercial questions still need evidence: do people finish useful work, return to it and pay repeatedly?
The next measures are activation, time to useful output, acceptance and repeat usage. Those behaviours should decide what we build next.
Explore Ghostpen ↗