← Alpha Horizon

Horizon One · Live Demo

Public simulations

Explore demo scenarios showing how decisions can ripple through markets, audiences and organisations.

Available simulations

📖

Simulation Tutorial

Interactive Guide to the Insights Portal

Welcome to the Alpha Horizon Insights Portal. This portal visualises the results of agent-based simulations, showing how synthetic consumer personas form and shift opinions about products and brands.

How to use

1. Select a simulation from the sidebar on the left.

2. Navigate tabs at the top to explore different visualisations.

3. Interact with charts by clicking, hovering, and using controls below each chart.

4. The header bar shows the total ticks (simulation time steps) and the last update date.

The Summary tab displays the Strategic Analysis Report — an LLM-generated markdown document that synthesises key findings from the entire simulation run.

What you'll find

• Executive Overview — High-level summary of sentiment shifts and key trends.

• Archetype Behavioural Analysis — How each consumer archetype reacted to the product/campaign.

• Influence Network Insights — Who were the most impactful influencers and how did peer pressure shape opinions.

• Strategic Recommendations — Actionable takeaways for marketing, product positioning, or audience targeting.

💡 This report is auto-generated by the LLM based on the simulation data. It adjusts the time-unit (Days/Weeks) to match the simulation parameters.

The Narrative Flow uses a Chord Diagram to visualise how different narratives and themes propagated across agent archetypes during the simulation.

How to read it

• Arcs around the circle represent different archetypes or narrative clusters.

• Chords (ribbons) connecting two arcs show the flow of narrative influence between them.

• Thicker ribbons = stronger narrative connection or influence flow.

• Hover over any ribbon to see the exact source, target, and strength value.

💡 This tab helps you understand which archetypes traded narrative influence and how stories spread through the simulated population.

The 3D Ripple Network is an immersive Three.js visualisation of the agent influence network. Each node is an agent, and edges represent influence connections.

Navigation Controls

W / A / S / DMove Forward / Left / Back / Right SPACEFly Upward SHIFTFly Downward MOUSELook Around (click canvas first) SCROLLAdjust flight speed

Features

• Nodes are coloured by archetype. Larger nodes = higher influence.

• Edges (green/white lines) represent influence connections between agents.

• Proximity indicator — fly close to a node to see its name appear at the top.

• Agent Inspector — use the dropdown below the visualisation to select an agent. The camera will fly to their position automatically.

The Impact Hierarchy displays a Sunburst Chart showing the influence chain radiating from a central agent.

How to read it

• Center circle = the primary influencer (the agent whose influence chain is displayed).

• Level 1 (Direct) ring = agents directly influenced by the center agent.

• Level 2 (Secondary) ring = agents indirectly influenced (influenced by Level 1 agents).

• Click on any segment to zoom in. Click the center to zoom back out.

Agent Impact Drill-Down

Use the Agent Impact Drill-Down dropdown below the chart to select any agent. The sunburst will completely rebuild around the chosen agent as the new center, showing their influence chain — who they directly and indirectly influence throughout the simulation.

Select "All Agents (Full Hierarchy)" to return to the default primary influencer view.

The Sentiment tab shows how agent opinions evolved over time, broken down by archetype and overall.

Charts Explained

• Opinion Timeline — Line chart showing the average opinion per archetype across each simulation tick (time step). The Y-axis is the opinion score (typically 0–1).

• Final Opinion Distribution — Bar or violin chart showing how opinions are distributed at the end of the simulation.

Key Metrics

final_sentiment — The average opinion at the end of the simulation (0 = very negative, 1 = very positive).

sentiment_change — The shift in average opinion from the start to the end of the simulation. Positive = opinions improved; negative = opinions worsened.

polarization_index — Measures how divided the population is. Values near 0 = consensus (everyone agrees); values near 1 = extreme polarisation (strong opposing camps).

min_opinion — The lowest individual opinion score recorded among all agents at the final tick.

max_opinion — The highest individual opinion score recorded among all agents at the final tick.

💡 Hover over chart lines to see exact opinion values at each tick. Use the Plotly toolbar (top right of chart) to zoom, pan, or download the chart as a PNG.

The Archetypes tab breaks down the simulation population into consumer archetypes — predefined persona types with distinct behavioural profiles.

Charts Explained

• Archetype Distribution — Pie or bar chart showing how many agents belong to each archetype.

• Opinion by Archetype — Grouped bar chart comparing average final opinions across archetypes. Helps identify which consumer types responded most positively or negatively.

What are archetypes?

Archetypes represent consumer personas with distinct traits — for example, a "Pragmatic Family Manager" values affordability and practicality, while an "Aspirational Upgrader" is trend-driven and brand-conscious.

Each archetype responds differently to media influence, peer pressure, and product features based on their programmed behavioural profile (e.g., openness to new ideas, resistance to peer pressure, media sensitivity).

The Agent Inspector appears below the 3D Ripple Network and allows you to deep-dive into any individual agent's behaviour and decision-making.

Inspector fields

• Agent Archetype — The consumer persona type of the selected agent.

• Live Opinion — The agent's final opinion score (0.000 to 1.000).

• Reasoning History — A tick-by-tick log of how the agent's opinion changed and why. Each entry shows the tick number, the opinion value at that point, and the reasoning text explaining the agent's thought process.

• Influence Capability — A table listing which other agents influenced this agent, the type of influence (Peer, Media, Direct), and the weight (strength) of that influence.

💡 In the 3D network, selecting an agent from the Inspector dropdown will also fly the camera to their node position.

• Plotly Chart Toolbar — Hover over any Plotly chart to reveal the toolbar (top right). Use it to zoom, pan, auto-scale, save as PNG, or reset the view.

• Switch simulations — Click any other simulation in the sidebar to instantly switch. All tabs will update to show data for the new run.

• Compare runs — Look at sentiment_change and polarization_index across different simulation runs to compare the impact of different parameters or scenarios.

• Screenshot — Use the 📷 camera icon (top right of charts) to capture the current chart view.

• Exit — Click the EXIT button in the header bar to return to the main homepage.

Back to site
Last Update
-
Node Name
WASD - Fly | SPACE/SHIFT - Up/Down | MOUSE - Look

Agent Profile

Unknown
Live Opinion
0.000
Reasoning History
Influence Capability
Target Type Weight
H1

HORIZON ONE

Admin Console

System Status

OPERATIONAL
Active Tenants
5
Total Sims
23
Compute Load
42%

Tenant Directory

Organization Folder Name Status Actions

Upload Simulation

Upload a simulation export JSON to any tenant.

Platform Administrators

Username Actions

Deploy New Tenant

Manage Tenant

Upload Simulation Run

Drop JSON, MD, or PDF here or click to browse

Add Admin Account

Update Administrator

Manage Simulations

Simulation ID Name Date Actions

No simulations found for this tenant.