JamaSoft Concept was exploring AI-first product bets. VideoHQ.ai was the most promising: a tool that helped non-experts produce structured, on-brand video content with AI.
Early AI video tools were either toys or required heavy editing skill. There was a clear gap for opinionated, workflow-driven AI video for operators and small teams.
- Define and launch a focused MVP
- Acquire the first 1,000 users
- Validate willingness to pay for AI-assisted video
- Establish a repeatable prompt-engineering practice
Ran problem interviews with marketers, educators, and creators. Prototyped five candidate workflows and pressure-tested them against real briefs.
Picked the three workflows with the highest pull and shipped them as a single product surface. Treated prompts as product artifacts with versioning and evals.
- Defined the V1 scope and prompt-engineering guardrails
- Partnered with engineering on the AI orchestration layer
- Designed the editor and review experience
- Drove launch positioning and acquisition channels
Keeping AI output quality consistent enough to trust, while moving fast enough to compete in a rapidly shifting AI-tool market.
VideoHQ.ai acquired 1,000+ users post-launch, validated the core willingness-to-pay hypothesis, and produced an internal AI product playbook reused on later launches.
- AI products live or die on output quality invest in evals from day one
- Pick workflows, not features. Opinion is the product
- Prompt versioning belongs in the product lifecycle, not in a doc