The work around content generation
A founder might have a useful conversation, a consultant might have years of material in documents and recordings, and a team might already have the idea they want to share.
The difficult work is finding the useful argument, checking claims, keeping the person’s voice, adapting it for different channels and returning to it later. I focused Ghostpen’s direction on that connected workflow.
- 01SourceBring existing knowledge
- 02DirectionChoose audience and purpose
- 03ProductionCreate the useful pieces
- 04ReviewEdit and approve
- 05ContinuitySave, return and reuse
My role in defining product direction
Ghostpen already had capabilities across capture, generation, voice personalisation, editing, review, saved work, mobile and collaboration. Almost any new feature could sound reasonable.
As Product Lead, I worked across product definition, scope, roadmap, interaction design and delivery. I wanted a clearer answer to who needed the product and what they needed it to do well.
Separating built capabilities from customer value
I separated what was built, what was partially working, what belonged in the target experience and what should wait. A screen or an API existing did not mean the whole journey was ready.
We narrowed the initial customer hypothesis towards independent experts, founder-led businesses and small professional-services teams. They already had valuable knowledge, but limited time to turn it into consistent content. That remained a hypothesis requiring customer evidence.
Source-led workflows and intentional output packages
I made the source central to the workflow. A recording, meeting, document, link or voice note already holds context; the user should not have to recreate it in a blank prompt.
The product direction moved towards intentional packages rather than maximum output count. That meant fewer visible options, but a clearer path to work someone could actually finish, review and use.
Editable AI work and explicit approval
Users needed to correct how Ghostpen understood the source, edit generated work and approve the result. Fluent output could still misinterpret a claim or lose the person’s voice.
AI could recommend and generate. Important edits, approvals, publishing and consequential changes still needed the user’s control. I used editable work, clear states and explicit approval to support that boundary.
A shared model for saved and reusable work
I organised the product around sources, campaigns, artifacts and decisions. That gave web, mobile and backend a shared way to describe where information came from, what the work was for and what the user had changed or approved.
Continuity mattered within a campaign, across sessions and in future work. If the argument changed, related outputs needed to stay aligned. When a user returned, they needed to see what was done and what still needed attention.
Setting boundaries for mobile and collaboration
I focused mobile on capture, lightweight framing, review, approval and returning to saved work. Deeper production could stay on a larger screen.
Room supported collaboration around Ghostpen’s work: contributing context, reviewing and resolving decisions. Keeping that boundary helped us avoid turning collaboration into a separate workplace messaging product.
We deferred full mobile parity, a full video editor, autonomous publishing and large enterprise administration before demand existed. The question was why a feature should be built now.
Delivery and authenticated staging validation
I worked across design, web, mobile, backend and QA to align product decisions with what the system could reliably support.
Validation covered authentication, uploads, source processing, generation, editing, saved work, state changes, interruptions, approval and error recovery.
These checks support technical readiness within the tested paths. They do not establish product-market fit.
Product outcomes and commercial validation
The clearest outcomes were a shared product model and a more disciplined roadmap. We had a better way to decide what belonged in the experience and what needed to wait.
I defined success around getting to a useful output, acceptance, completion, returning to saved work and repeat usage. These measures needed to show whether customers valued the workflow enough to keep using and paying for it.
What I learned about prioritisation
Ghostpen made me spend more time on whether we should build something before deciding how to design it.
Technical capability gave us many possible directions. My job was to keep the customer problem clear, build enough to test the assumption and let behaviour guide what came next.
Explore Ghostpen ↗