I Built an AI Simulation Engine You Talk to Like ChatGPT — Meet MiroFish

ce zhang — Wed, 06 May 2026 15:19:42 +0000

What if you could rehearse a crisis before it happens, pressure-test a campaign before you spend, or map the second-order reactions to a pricing change — all by asking a question in plain language?

That's MiroFish.

The Problem: Why ChatGPT Isn't Enough

Ask ChatGPT for a prediction, and it collapses competing audience reactions into one confident answer. But real-world outcomes depend on people reacting to people — shifting incentives, emergent narratives, and resistance clusters that a single summary paragraph can't capture.

So I built something different.

Architecture: Multi-Agent Simulation in a Chat Interface

MiroFish runs a 5-stage pipeline behind a single chat conversation:

User Question → Knowledge Graph → Agent Simulation → Report → Follow-up Queries

Stage	What Happens
1. Seed	Plain-language question + optional file uploads (PDF, Markdown) for grounding
2. Graph	LLM extracts actors, relationships, pressures, and factual anchors into a structured knowledge graph
3. Simulate	AI personas interact across short-form and threaded social surfaces over multiple rounds — each with persistent memory and distinct incentives
4. Report	Structured result card with executive summary, risk signals, narrative paths, and confidence indicators
5. Query	Keep asking the simulated world — unlike a static forecast, this is interactive

The key insight: agents don't just answer. They react to each other, creating emergent behavior that a flat prompt-response loop can't produce.

What You Can Use It For

Use Case	Example
Campaign Testing	"What happens if we launch in a skeptical category?"
Pricing Reactions	"If we raise prices, which segments push back first?"
Policy Stress Tests	"Where does support split when this goes public?"
Market Narratives	"What if positive news meets coordinated skepticism?"

Writing a Good Prompt

Name the decision, the audience, the trigger, and the time horizon:

❌ "What happens if we change pricing?"

✅ "What happens to customer trust if we remove the bundled charger from the flagship model next quarter?"

Try It

No signup. No setup. Ask a question.

👉 mirofish.homes

Open source on GitHub.

Introducing MiroFish: Predict Anything, But Talk to It Like ChatGPT

ce zhang — Wed, 06 May 2026 15:01:19 +0000

What if you could rehearse a crisis before it happens, pressure-test a campaign before you spend, or map the second-order reactions to a pricing change — all by asking a question in plain language?

That's MiroFish.

MiroFish is an AI simulation engine that lets you talk to a scenario the way you'd talk to ChatGPT. Ask a question. Upload a document if you want. Behind the scenes, a multi-agent system builds a knowledge graph, runs agent-based simulations across social surfaces, and delivers a structured prediction report — all inside a single conversation.

Why Not Just Ask ChatGPT?

Single-model answers collapse competing audience reactions into one confident response. MiroFish doesn't.

It creates a living scenario: agents with distinct personas, incentives, and memories interact over multiple rounds. You get to watch narrative spread, resistance clusters, and emergent behavior — not a single summary paragraph.

Approach	What you get
Single chat answer	Fast, useful, but often one-dimensional
Manual research	Thorough, but slow when many groups interact at once
MiroFish	Agents, memory, social surfaces, and a report you can keep questioning

How It Works

Seed — Start with a plain-language question. Add a strategy memo, policy brief, or customer research as optional grounding.
Graph — The engine extracts actors, relationships, pressures, and factual anchors into a structured knowledge graph.
Simulate — Personas interact across short-form and threaded social surfaces over multiple rounds.
Report — A prediction report surfaces turning points, risk signals, narrative paths, and confidence indicators.
Keep Asking — Unlike a static forecast, you continue questioning the generated world to explore counterfactuals.

What Can You Use It For?

🎯 Campaign Testing

"What happens if we launch this positioning in a skeptical category?"
Simulate how audience groups might amplify, resist, or reinterpret your message before you commit budget.

💰 Pricing Reactions

"If we raise prices next quarter, which customer groups push back first?"
Model sentiment, value perception, and likely objection paths across different segments.

🏛️ Policy Stress Tests

"If this policy enters public debate, where does support split?"
Use simulation as a tabletop exercise for controversy, coalition formation, and second-order effects.

📈 Market Narratives

"What if positive news meets coordinated skepticism on social channels?"
Stress-test market stories where spreadsheets miss the feedback loop between analysts, retail attention, and public discourse.

What Makes a Good Prompt?

Name the decision, the audience, the likely trigger, and the time horizon. A narrow question gives the simulated world less room to drift.

❌ "What will happen if we change our pricing?"

✅ "What happens to customer trust if we remove a bundled charger from the flagship model next quarter?"

The Report You'll Get

Every answer drops a structured result card with:

Executive summary — likely trajectory at a glance
Risk signals — what could derail the outcome
Narrative paths — how the story spreads (and where it fractures)
Follow-up questions — "Which persona creates the first negative cascade? What changes if we announce a transition plan first?"

Is This a Guaranteed Forecast?

No. And it doesn't pretend to be.

MiroFish is exploratory decision support — a way to rehearse plausible reactions, surface blind spots, and sharpen your own judgment before you use analytics and real-world validation.

Try It

If you're planning a launch, testing a pricing change, or staring at a policy draft wondering what you're missing — open MiroFish and ask it a question.

No setup required. Start with text, add files when you want more grounding.

👉 mirofish.homes

DEV Community: ce zhang

I Built an AI Simulation Engine You Talk to Like ChatGPT — Meet MiroFish

The Problem: Why ChatGPT Isn't Enough

Architecture: Multi-Agent Simulation in a Chat Interface

What You Can Use It For

Writing a Good Prompt

Try It

Introducing MiroFish: Predict Anything, But Talk to It Like ChatGPT

How It Works

What Can You Use It For?

🎯 Campaign Testing

💰 Pricing Reactions

🏛️ Policy Stress Tests

📈 Market Narratives

What Makes a Good Prompt?

The Report You'll Get

Is This a Guaranteed Forecast?

Try It