Room

Your pocket board of directors. A minimalist AI team for solopreneurs.

Comment

Inspiration

Solopreneurship is weird. You have the freedom of a CEO, but the resources of an intern. The biggest pain point isn't lack of ideas it's isolation and chaos.

When building products, creators often get stuck in "tinkering loops" (tweaking colors instead of shipping) or "validation vacuums" (building things nobody wants). We don't need another generic "cheerleader" chatbot saying "You can do it!". We need a rigorous system. We need a Board of Directors.

Inspired by the philosophy of "Systems over Motivation" (popularized by creators like Simon from Better Creating), I built Room. I wanted to replace the messy, amnesiac chat history of standard LLMs with a persistent, context-aware team of experts that lives in your pocket.

What it does

Room is a minimalist AI workspace that acts as your personal Board of Directors. It is designed for zero-friction coaching.

  • Zero-Friction Onboarding: You don't sign up with email/password. You simply tell Room:
  • Who you are.
  • What you are building (e.g., "A Notion OS for Agencies").

  • Your biggest constraint (e.g., "Stuck tweaking the design").

  • The Board (AI Personas): Instead of one generic bot, you have access to specialized agents:

  • The Generalist (Free): Your versatile operations manager.

  • The Strategist (Pro): A systems thinker who focuses on workflows and roadmaps.

  • The Critic (Pro): A "Devil's Advocate" who rigorously stress-tests your ideas and pricing.

  • The Mentor (Pro): Focuses on founder psychology and resilience.

  • Context Injection: Every message you send is silently wrapped with your project context. You never have to repeat "I am building a SaaS..." again. The AI knows.

How I built it

I built Room with a focus on speed ("Speedrun Mode") and radical minimalism.

  • Framework: Flutter & GetX. I chose Flutter for the pixel-perfect UI and GetX for rapid state management and dependency injection.
  • AI Engine: Z.AI GLM-4.7 API. I engineered specific system prompts for each persona to ensure they sound distinct (e.g., The Critic is instructed to be harsh but constructive).
  • The "Context Equation": To ensure high-quality responses, I structured the prompt injection mathematically:

Where includes the user's current project and active constraints.

  • Monetization: RevenueCat. I implemented a seamless paywall that locks specific "Pro" personas (The Strategist, The Critic). The integration was handled using the purchases_flutter package, allowing me to ship a working subscription model in under 2 hours.

Challenges I ran into

  • Prompt Engineering vs. Personality: Making "The Critic" sound helpful rather than just mean was a challenge. I had to fine-tune the system instructions to ensure the feedback was actionable (e.g., "Critique the pricing, don't insult the user").
  • The "Blank Canvas" Problem: Initially, the chat screen was too empty. I solved this by implementing "Suggestion Chips" and an auto-greeting system that uses the user's name and project data to initiate the conversation, effectively killing writer's block.
  • Time Constraints: Building a polished, bug-free UI with complex state management + monetization in a hackathon timeframe required cutting many "nice-to-have" features (like voice input) to focus purely on the core text experience.

Accomplishments that I'm proud of

  • The "Pro" Feel: I resisted the urge to use colorful gradients. Sticking to a strict Black & White (#000000 & #FFFFFF) palette gives the app a premium, "Notion-like" aesthetic that fits my target audience perfectly.
  • RevenueCat Integration: I successfully implemented a full paywall flow from the "Lock Icon" trigger to the successful purchase listener which instantly unlocks the premium AI agents.
  • Immersive Audio (TTS): I didn't stop at text. I implemented Text-to-Speech capabilities, allowing the "Board of Directors" to speak their advice out loud.
  • Context Persistence: The app genuinely feels like it "knows" you. Switching from The Generalist to The Strategist feels like walking from one department to another in a real company, carrying your project file with you.

What I learned

  • Context is King: The difference between a "Chatbot" and a "Coach" is simply memory. By hard-coding the user's constraints into the prompt, the AI becomes 10x more valuable.
  • Monetization as a Feature: Using RevenueCat taught me that a paywall isn't just a barrier; it's a signal of value. Locking the "best" agents makes the user perceive them as higher utility.
  • Flutter is Fast: Using GetX with Flutter allowed me to iterate on UI changes in seconds, which is crucial for a 48-hour sprint.

What's next for Room

  • Two-Way Voice Conversation: Currently, the AI speaks to you (TTS). Next, I want to implement Voice Input (STT) so founders can have full verbal board meetings while walking, without looking at a screen.
  • Integration: Connecting Room directly to Notion or Linear APIs to read/write tasks.
  • More Personas: Adding niche experts like "The Copywriter" or "The VC" (Venture Capitalist simulator).
  • Public Launch: Polishing the app for a real release on the Play Store and App Store.

Built With

Share this project:

Updates