“What if asking ChatGPT/Claude/Gemini wasn’t about getting the right answer — but watching the argument that gets you there?”
This is the question that caused me to launched GetIdea.ai — a side project that became a system, a system that became a mirror, and a mirror that occasionally throws insults at my ideas (thanks Harsh Critic 🔥).
Over the past few months, I’ve been building a multi-agent AI interface where ideas are tested not by a single voice, but by a council of distinct personalities. It’s not production-scale, but it’s already the most honest and useful AI interaction I’ve ever had. It started with a simple, frustrating problem:
🤔 The Problem: One Voice Is Almost Never Enough
First, if you’re at all like me you may feel most AI chats are really monologues disguised as a dialogue.
Even with all the right prompting, I still had to:
- Play the Devil’s Advocate
- Be the Strategic Thinker
- Remember the market context
- Question my own biases
And worst of all — I had to trust that the AI would play fair when it was really just playing along.
What I wanted wasn’t “help.” What I wanted was debate. Structured, selective, emotionally differentiated debate.
💡 The Concept: Assemble the Squad
So I built GetIdea.ai, a real-time multi-agent system where AI personas argue with each other so I don’t have to.
You ask a question — like:
“Should I quit my job to start an indie game studio?”
And instead of one fuzzy maybe-response, you get a brutal realist, a business strategist, and sometimes a Hype Champion trying to gas you up just enough to ignore them both.
What started as a test of CrewAI and WebSocket orchestration became a real product with:
- Confidence-based agent routing
- Conversation memory
- Typing indicators
- Real-time squad assembly
- Modular personalities like:
This isn’t just multiple answers — it’s structured dissent, with context, consistency, and enough personality to feel like a team, not a trivia night.
🧠 How It Actually Works (And Why It Matters)
Let’s say you ask:
“Should I quit my job to start an indie game studio?”
Here’s what happens under the hood:
🧠 GetIdea.ai instantly assembles a confidence-filtered squad of AI advisors. In this case:
- 🔥 Harsh Critic: high confidence in risk analysis, startup failure patterns, and tough love
- 💰 Business Strategist: high confidence in financial models, scalability, and unit economics
- 💡 Creative Catalyst: low confidence in risk — but very high in vision, innovation, and potential
Now here’s the kicker:
Each agent knows its strengths. It doesn’t pretend to be an expert in everything — it plays its role.
🎭 The Augment in Action (Yes, this really happened)
So, how does it work? Let’s use your question: “Should I quit my job to start an indie game studio?”
Instead of one answer, the system instantly assembles an advisory squad based on confidence. For this query, it routes to three distinct personas: a visionary, a pragmatist, and a realist.
Then, you get to watch the argument unfold in real-time.
🔥 Harsh Critic:
“The market is brutally competitive and often financially disastrous. Without groundbreaking innovation and a financial safety net, it’s a fool’s errand. Keep your job while testing your concept on the side.”
💰 Business Strategist:
“Consider your revenue model and target audience. Ensure diversified income streams and validate your concept with market research before making a decision.”
💡 Creative Catalyst (not triggered this time):
(Filtered out) — their confidence score wasn’t high enough to weigh in unless you lowered the threshold.
🔥 Harsh Critic:
“Sure, the Strategist makes good points — but you’re not in the business yet. Talking revenue models before you’ve validated demand is like planning a restaurant menu before knowing if anyone’s hungry.
🧪 Why That Matters
This isn’t a debate for entertainment. It’s a mirror for your own bias.
If you:
- Want optimism → lower the threshold, and 💡 Creative Catalyst joins.
- Want realism → leave it high, and only the tough love sticks.
- Want synthesis → add 🔄 Synthesis Master, who bridges their insights into a decision framework.
That’s the hidden power of GetIdea.ai: It’s not answering your question — It’s helping you ask better ones.
🧪 The Experiment Behind the Scenes
There’s a hidden slider in the UI: Confidence Threshold. Slide it down, and you get wild ideas. Slide it up, and only the most certain agents speak.
That single control taught me more about my own bias than I expected. If I don’t want to hear Harsh Critic, it’s not because he’s wrong — it’s because I’m not ready for him. But when I am ready? His hit rate is scary.
Also — each conversation starts with “assembling your expert advisory team.” Because that’s how this should feel: like you’re being heard, not processed.
✨ Why This Matters (to Me and Maybe to You)
This isn’t a startup pitch. Not yet.
But it’s a signal. That we’re moving from:
- Query → Answer to
- Question → Assembly → Synthesis
That’s not just more useful — it’s more human.
And honestly? It made me want to ask better questions.
👀 Coming Next in the Series
In Part 2: “The Build”, I’ll share:
- The architecture I’m modernizing
- Why crew_chat.py is 2,100 lines of chaos (and still worked)
- What went wrong (and hilariously right)
- How this system gave me real-time feedback on my own decision patterns
And eventually in Part 3: “The Payoff”, I’ll show where this is going — and why multi-agent systems might become the UI layer for better thought, not just better output.
✅ TL;DR (because I built this for people like me):
GetIdea.ai is:
- A real, working multi-agent chat system
- Built in CrewAI, FastAPI, and WebSocket magic
- Designed to simulate collaborative, conflicting, yet emotionally readable decision-making
- Still messy under the hood, but intentionally honest in tone
And maybe… it’s the future of how we talk to machines. By teaching them to talk to each other first.
🔗 Your Turn: Test It, Shape It, or Join In
The project is live, and this is where you come in. I’d be grateful for your help in any of these three ways:
- 🧪 Share Your Results: Try the tool with a real problem you’re facing. Post the most surprising or insightful piece of advice you get in the comments below.
- 💡 Suggest a Persona: What expert is missing from the council? A ‘Legal Advisor’? A ‘Marketing Guru’? Comment with the persona you think I should build next.
- 🤝 Become a Beta Tester: For those who want to go a step further, I’m looking for a handful of people for a 15-minute feedback session to help improve the experience. If you’re interested, just comment “I’m in!”
You can try the system right here: GetIdea.ai
I’m excited to hear what you think!
