Alice Kang
Los Angeles, CA
Blueprint AI
Independent artists had tools to release music, but very little support when it came to planning, pitching, or growth.
We saw an opportunity to treat guidance as a product, not a service.
Blueprint AI was designed to deliver personalized direction at scale while earning trust before monetization.
The problem, in plain terms
UnitedMasters had built distribution rails, but artists needed more than tools…they needed guidance.
- How do I pitch my next release?
- Where should I focus my energy?
- When do I drop, and what do I say?
We had a Music Team with deep knowledge, but no way to scale it. Reps could only help a fraction of our artists. Everyone else was guessing.
What made it hard
This was not a simple automation problem.
Several constraints shaped early decisions:
- The Assistants API we planned to use was still in beta, with limited reliability guarantees
- Artist metadata varied widely, from rich histories to sparse profiles
- Responses needed to feel personal and relevant, not generic
- Midway through development, API costs increased, forcing monetization questions earlier than planned
We could have reduced scope or leaned on templated advice. Instead, we prioritized trust and usefulness, even if that meant slower progress.
What we actually did
I was responsible for end to end product and design. That included:
- Defining the core interaction model, including onboarding, prompts, and feedback loops
- Designing a credit based monetization system once free usage at scale became unsustainable
- Creating prompt logic grounded in real artist data rather than static templates
- Partnering closely with product and engineering on weekly tradeoff decisions
- Supporting go to market assets with marketing so product behavior and messaging stayed aligned
- Synthesizing feedback from pilots, QA, and the Music Team to refine tone and scope
The goal was to make the tool immediately useful, even for artists with limited data.
How we rolled it out
We released Blueprint AI in phases to validate trust and value before broad exposure.
Phase 0: Internal QA (1st iteration)
Music Team reps reviewed outputs in detail. Their feedback shaped guardrails, fallback logic, and tone.
Phase 1: Targeted pilot
We launched with artists who had recently released music. Early prompts focused on release timing and playlist pitching.
"How it works" preview
Phase 2: Partner rollout
Access expanded to SELECT and PARTNER tiers across web, iOS, and Android. We monitored engagement and refined onboarding where confusion surfaced.
Artists see smart suggestions based on their upcoming releases, audience data, and real questions sourced from our Music Team.
Phase 3: Public launch and monetization
Once usage patterns confirmed value, we introduced pay as you go credit bundles. This helped manage costs without blocking early trust.
Purchasing credit bundles
Decisions that mattered
1. Prove value before monetizing
We delayed pricing until repeat usage showed clear intent. This reduced early revenue but protected credibility.
2. Anchor advice in real context
Every prompt was driven by the artist's actual data. Release history, timing signals, and performance patterns shaped responses.
3. Set clear boundaries
We surfaced examples, suggested questions, and clearly defined what the tool could and could not do.
Early signals
Final thoughts
This wasn't just an AI feature. It was a retention tool, a growth lever, and a trust challenge. We shipped something useful, fast, and human..while solving a real business constraint.
The metrics were strong, but the real win was access. Artists who never had support now have a coach in their pocket. They're planning smarter, pitching better, and building careers with help they didn't have before.
Blueprint AI shows that in creative industries, AI's value isn't replacement, it's reach. With the right design, data, and intent, it can democratize expertise and unlock progress for more people.