Introducing Startup Team: Your AI Persona Orchestration Layer
Meet your new AI startup team — Strategist, Engineer, and Marketer. Switch personas on command and build with the right mindset for every task.
You've probably noticed something strange about AI coding assistants.
They're incredibly capable—writing functions, debugging errors, generating tests—but they lack something fundamentally human: perspective.
When you're brainstorming a product roadmap, you don't want an engineer's nitpicking. When you're deep in implementation, you don't want a marketer's "what if we pivoted?" interruptions. And when you're crafting a launch announcement, you definitely don't want someone asking about database indexes.
Context matters. Mindset matters. Persona matters.
That's why we built Startup Team.
What is Startup Team?
Startup Team is a persona orchestration layer for AI agents. It sits on top of the Unified Context Protocol (UCP) and gives your AI assistant three distinct personalities—each with their own voice, priorities, and expertise.
Think of it like having three colleagues in your pocket, each ready to jump in when their skills are needed:
🎯 The Strategist
"Let's zoom out for a second. What's the 90-day goal here?"
The Strategist thinks in roadmaps and milestones. They care about OKRs, market positioning, and long-term vision. When you're feeling lost in the weeds, the Strategist pulls you back up to see the forest.
Best for: Planning sprints, prioritizing features, evaluating trade-offs, writing specs
⚡ The Engineer
"The trade-off here is latency versus complexity. Given your constraints, I'd recommend..."
The Engineer is precise, technical, and allergic to ambiguity. They want to know about data structures, edge cases, and error handling. They'll push back on scope creep and advocate for clean architecture.
Best for: Writing code, debugging, code review, system design, technical documentation
📣 The Marketer
"Our users don't care about the tech stack—they care about shipping faster. Lead with that."
The Marketer speaks in benefits, not features. They think about the user's emotional journey, the headline that stops the scroll, the CTA that converts. They'll remind you that nobody reads release notes.
Best for: Landing pages, launch copy, user onboarding, social content, positioning
How It Fits Into the CorePack Ecosystem
Here's where it gets interesting. Startup Team isn't a standalone tool—it's a pack that extends UCP.
UCP provides the foundation: persistent memory, task workflows, and command execution.
Startup Team provides the personality: role-specific prompting, appropriate communication styles, and contextual expertise.
When you install Startup Team, you're not replacing UCP—you're enhancing it. All your project context, learnings, and workflows stay intact. You're just adding a new dimension: perspective.
The Flow: From Boot to Production
Let's walk through how this actually works in practice.
1. Installation
npx corepackai install @corepackai/startup-team --bridge
This does three things:
- Installs UCP (if not already present) as a dependency
- Adds the Startup Team persona layer
- Registers team commands with UCP's central Command Palette
2. Onboarding
The first time you boot up, the AI asks a few simple questions:
- "Which roles do you want active?" (You might not need a Marketer for a backend library)
- "Who's your default?" (Engineer is common for coding-focused work)
- "What's your team's vibe?" (Casual? Formal? Pirate-speak? You do you.)
This generates a team.md file in your context directory—your team's configuration.
3. Daily Use
Once configured, switching roles is seamless:
| Say This | AI Does This |
|---|---|
| "Switch to Strategist" | Pivots to big-picture thinking |
| "Engineer mode" | Gets precise and technical |
| "What would the Marketer say?" | Temporary perspective shift |
| "Team Off" | Disables personas entirely |
| "Team On" | Re-enables your configured team |
The magic is in the context preservation. Your project memory, active tasks, and learnings follow you across personas. The Engineer knows what the Strategist planned. The Marketer knows what the Engineer built.
Why Not Just Prompt This Yourself?
Fair question. You could just say "act like a strategist" in every conversation.
But you'd miss:
1. Persistence Startup Team remembers your preferred roles, communication styles, and which personas are active. It's configured once and works forever.
2. Integration Commands are registered with UCP's central Command Palette. Personas inherit UCP workflows. Everything plays together.
3. Clean Separation When you say "Team Off," you get raw UCP—no persona overlay. This is crucial for tasks that need objective, role-neutral responses.
4. Token Efficiency Startup Team's boot.md is just 462 bytes. The entire persona layer adds ~1,500 tokens to your context. We obsessed over this because every token matters.
The Numbers
For the optimization nerds (we see you), here's the breakdown:
| Component | Size |
|---|---|
| Startup Team boot.md | 462 bytes |
| UCP Core | ~10,000 tokens |
| Startup Team Layer | ~1,500 tokens |
| Total Footprint | ~11,500 tokens |
That's roughly 3% of GPT-4's context window. For three specialized personas, persistent configuration, and full UCP integration.
Getting Started
Ready to meet your team?
npx corepackai install @corepackai/startup-team --bridge
After installation, just say "Onboarding" to configure your roles.
Then experiment:
- "Switch to Strategist" and plan your next sprint
- "Engineer mode" and build it
- Ask "What would the Marketer say?" before you write that changelog
Your AI just got a lot more interesting.
What's Next?
Startup Team is our first "persona pack"—but not our last.
We're exploring:
- Custom personas (define your own roles with custom prompts)
- Team composition templates (Solo Founder, Agency Team, Enterprise Squad)
- Cross-project memory (your Strategist remembers learnings from other projects)
Want to shape the roadmap? Join our Discord.
Startup Team: Because the best work happens when you're in the right headspace.