Simulate your launch before your customers react to it

Write your launch brief

The simulation can only react to what you tell it. A vague brief produces a vague crowd, so put the effort in here.

1. Open a blank document and copy the template below into it.
2. Fill in every field with plain, factual sentences. Write what you would say on the sales page.
3. Add your real weak spots. The crowd finds them anyway, and naming them gives the simulation something to test.
4. List the two or three competitors or alternatives your buyers already use.
5. Save the file as launch-brief.md or launch-brief.pdf.

LAUNCH BRIEF

Product: [name]
What it is: [one plain sentence]
Who it is for: [role, business size, situation]
Price and offer: [price, bonuses, guarantee, deadline]
Launch date and channels: [date, where you will announce it]

Key claims we will make:
- [claim 1]
- [claim 2]
- [claim 3]

Known weak spots:
- [what a skeptic would point at first]

Competitors and alternatives:
- [name and what it does better or worse]

Questions we expect from buyers:
- [question 1]
- [question 2]

Sign in to MiroFish

1. Open the MiroFish homepage in your browser.
2. Click Sign In in the top right corner.
3. Finish the sign-in flow. The console button on the homepage reads Sign in to start engine until you do.

Upload your brief

1. Find the panel labelled Reality Seeds in the engine console.
2. Choose Upload File, or drag your file into the drop area.
3. Confirm the file type is PDF, MD, Markdown, or TXT and that the total size is under 50MB.
4. If your launch details live on a web page, use Import Link instead.

Add your simulation prompt

The prompt tells the crowd what to focus on. One clear question beats a list of five.

1. Find the panel labelled Simulation Prompt.
2. Paste one of the prompts below. Start with the first one on your first run.
3. Replace the bracketed parts with your own details.

Launch stress test. This is the prompt MiroFish suggests for product launches, open betas, and brand announcements.

Simulate how customers, competitors, and commentators react to this
launch brief, and identify the most expensive misunderstanding if left
unanswered.

Price test. Use this when you are unsure about your number.

Simulate how potential buyers and skeptics react to the price of
[price] in this launch brief over three rounds. Show which objections
appear first, which ones spread, and what proof would change their minds.

Competitor comparison. Use this when buyers already have an alternative.

Simulate how buyers compare this launch with [competitor]. Identify
which claims they believe, which they doubt, and which comparison
gets the largest group to switch.

Check the graph before you run

MiroFish first extracts the people and pressures in your brief. This is your chance to catch a misread.

1. Start the engine. MiroFish moves through ontology generation and graph construction.
2. Open the graph view and look at the actors it found: buyers, competitors, commentators, and any groups you named.
3. Check that your main audience is present and that the tensions match your real concerns, such as price against value.
4. If a key actor or concern is missing, edit your brief to state it directly, then upload the new version.

Run the simulation

1. Start the parallel simulation from the console.
2. Watch the feed. Agents post, reply, argue, and shift their opinions across Twitter-style and Reddit-style surfaces.
3. Note the first objection that appears and who raises it.
4. Note any reaction that spreads quickly, even when it is wrong. A wrong idea that travels is a messaging problem you can fix before launch.
5. Wait for all rounds to finish before you move on.

Read the prediction report

The report is the deliverable. Read it with a pen and mark every claim you can check.

1. Open the report view once the simulation completes.
2. Read the executive summary first and write down the dominant risk in one sentence.
3. Scan the predicted developments, major risks, key actors, and evidence lines.
4. Copy every objection and misunderstanding into a single list, ordered by how early it appeared and how far it spread.
5. Mark the confidence signals. Where confidence is low, treat the finding as a question, not an answer.

Turn objections into tests

This is where the simulation pays off. Each objection becomes a small experiment with real customers.

1. Use the deep interaction step to question the ReportAgent. Ask why a specific objection spread.
2. Interview individual characters. Ask a skeptical buyer what single change would earn their trust.
3. Take your objection list and match each item to a test in the table below.
4. Run the tests with five to ten real people from your audience before you launch.
5. Update your brief with what you learn, then run a second simulation to see if the reaction shifts.