Stack

The scope was unusually wide for a first product. Rather than a focused MVP around one workflow, the founding bet was that consolidation is the value proposition, so a thin version of the whole surface is worth more than a deep version of one part. That meant designing 12 or more core modules end to end, from onboarding and dashboard through course building, recording, document and presentation authoring, marketplace and monetisation, team management, and account security.

Two further problems shaped structure rather than features. Because assets are produced in tools with different templates and typefaces, brand consistency across certificates, landing pages, and community spaces is effectively impossible to hold. And as a creator grows into a small team, roles and permissions have to be configured independently in every one of those tools, with calendars and communication living in yet more.

The design constraint on top of all this was that one person had to cover the entire surface. With a solo designer and a startup timeline, no module could afford a bespoke interaction language, because the cost of inconsistency across 12 modules would land on the engineering team as 12 separate builds.

**Sparky as a layer, not a destination.** The AI assistant is reachable from inside every module and generates in the format of the module it is called from: course structures in the builder, teleprompter scripts in the recording suite, body copy in the document editor, slides in the presentation builder. The assistant follows the work instead of the work travelling to the assistant.

**A shared brand kit.** Logos, colours, fonts, and templates are defined once and applied automatically across documents, presentations, certificates, and storefront pages, which turns brand consistency from a discipline the creator has to maintain into a property of the system.

On top of those layers sit the modules themselves: a guided onboarding flow that captures goals and tailors the dashboard from the first session, a home dashboard that opens directly into creation, an AI course builder that generates a full module and lesson framework from a single prompt and supports video, text, assignments, and a certificate designer, an in-browser recording suite with screen and camera capture, an AI teleprompter, scene management, and export up to 4K, a document and presentation creator with AI generation and granular sharing, a dual-sided marketplace serving creators as sellers and students as buyers with flexible pricing models, a business suite covering team roles, shared calendar, messaging, and Stripe, PayPal, and Google Meet integrations, and a security and settings area handling two-factor authentication and a 30 day grace period on account deletion.

Adopting Align UI as the foundation rather than building a design system from scratch is what made that breadth achievable. It supplied a component architecture from day one, so the work went into product thinking and module structure instead of into redrawing inputs and menus 12 times.

Adopting an existing design system was the highest leverage decision on the project and the one I was most reluctant about. The instinct with a new product is to build its language from scratch, and I had to weigh that against the schedule honestly. Taking Align UI meant the interface was not fully original, but it bought back the time that went into the product thinking the founding team actually needed, and it made consistency structural instead of a thing I had to enforce by hand across 33 sections.