AI Tooling for Game Teams — Early Access

AI Tooling for
Game Teams

We build NPC dialogue workflows and prototyping tools for game developers. Currently in early access — focused on making AI-powered NPCs accessible to indie teams.

Try the Demo Learn More

Built for game development workflows

GCP

Alibaba Cloud

LLM-Powered NPCs

Multi-Language

Technology We Use

GCP

Cloud Platform

Alibaba Cloud

Cloud Services

SQLite

Local Persistence

Node.js

Express API Server

TLS / HTTPS

Encrypted Transit

Early Access

Current Stage

NPC Dialogue

Primary Focus

Teams building games like these use our tools

Freebird Games

To the Moon

Narrative RPG

Red Thread Games

Dreamfall Chapters

Narrative Adventure

Stoic Studio

The Banner Saga

Tactical RPG

Spiderweb Software

Avernum, Geneforge

Classic CRPG

Wadjet Eye Games

Blackwell Series

Point-and-Click

Kitfox Games

Boyfriend Dungeon

Sim / Dungeon Crawler

Gareth D. Martin

Citizen Sleeper

Sci-Fi Narrative RPG

Soldak Entertainment

Din's Curse

Dynamic World RPG

Kiro'o Games

Aurion

Action RPG

CBE Software

J.U.L.I.A.

Sci-Fi Adventure

What We Build

AI Tooling for Game Teams.

Vantage Digital Labs builds AI tooling for game teams, focused on NPC dialogue workflows and prototyping. We’re a small engineering team working to make AI-powered NPCs accessible to indie developers — not just studios with deep pockets.

Our current focus is a cognitive NPC engine that powers dynamic, in-character dialogue with emotional state tracking and multi-language support. The demo on this page is powered by a live inference backend — try it out.

Contact

Current stage: Early access. We’re actively developing the NPC dialogue engine and iterating based on feedback from early users.

About Us

Vantage Digital Labs is a small, focused engineering team building AI tooling for game studios. We believe cognitive NPCs should be accessible to indie teams, not just enterprises with deep pockets. Founded by Murni Marcus.

Our codebase is developed in the open. Check out our SDKs, report issues, or contribute on GitHub. For partnership inquiries or support, reach us by email.

GitHub Repository Contact Us

What We're Building

AI Modules for Game Teams

Focused on NPC dialogue workflows and prototyping tools.

Active

Cognitive NPCs

Dynamic NPC dialogue with emotional state tracking and multi-language support. Powers the live demo on this page.

Try the Demo →

Research

Procedural World Gen

Generating 3D assets and level layouts from text prompts. This is early-stage research — not yet available.

Not Yet Available

Research

Anti-Cheat Analysis

Behavioral telemetry analysis for detecting suspicious player patterns. Early-stage research concept.

Not Yet Available

How It Works

The NPC demo on this page connects to a live inference backend. Here's a simplified view of the architecture.

Inference Proxy

Server-side API proxy that routes requests to LLM providers. Provider details are hidden from the frontend — only the Vantage API is exposed.

NPC System Prompts

Each NPC has a defined personality, backstory, and rules. The LLM generates contextual, in-character responses with emotion tracking.

Language Detection

NPCs automatically detect the player's language and respond in the same language. Supports any language the underlying LLM can handle.

Experience Cognitive NPCs

Try our browser-based Sandbox Simulator below.

Operator-7

ORBITAL STATION GUARD

Sentiment Suspicious

HostileNeutralFriendly

Trust Level

12%

Alert Level

MEDIUM

Engine:Active

Model:Vantage LLM v2

Region:asia-northeast1

Exchanges:0

Terminal Connection_

Live AI Backend

OP-7 [System]

Halt. This sector of the orbital ring is restricted. State your clearance code immediately, or I'll be forced to alert the sentries.

How This Demo Works

This demo connects to a live AI backend. Operator-7 is an NPC with a defined personality, backstory, and emotional state — responses are generated in real-time by the language model, not from pre-written scripts.

1

Type a message in the terminal input and press Enter or click the send button.

2

Your message is sent to our inference API. The NPC generates contextual, in-character responses based on its system prompt personality.

3

Watch the Sentiment indicator on the left panel change based on the conversation — it shifts between Neutral, Suspicious, Hostile, Helpful, etc.

4

In production, the NPC engine maintains conversation history and emotional state across sessions, with support for multiple languages.

Try These Example Inputs

Click any example below to auto-fill and send it to the NPC:

Tip: The NPC can respond to any topic — try asking about the station, its secrets, or just have a conversation. The sentiment meter reflects the NPC's emotional state.

Disclaimer: This demo uses a live AI backend for dialogue generation. Responses are generated in real-time by a language model — they are not pre-written scripts.

Transparent Pricing

From prototype to production scale. Volume discounts for high-throughput studios.

Sandbox

For prototyping and game jams. No credit card required.

$0/mo

Free forever

5,000 NPC inference calls / month
1 NPC personality slot
500 PCG mesh generations
Community Discord support
Sandbox environment only
No production deployment
No TTS voice synthesis

Pro

For indie studios shipping to production.

$79/mo

Billed annually ($99/mo if monthly)

100k NPC inference calls / month
10 NPC personality slots
5,000 PCG mesh generations
TTS voice synthesis (5 voices)
Episodic memory enabled
Production API keys
Email support (48h SLA)
No Anti-Cheat module

Enterprise

For studios with high CCU or custom requirements.

Custom

Starting at $499/mo, volume discounts available

Unlimited NPC inference calls
Unlimited personality slots
Unlimited PCG generations
TTS with custom voice cloning
Anti-Cheat Sentinel module
Dedicated GPU cluster
Priority support
Slack/Discord priority support (4h SLA)

Usage-Based Overage Pricing

NPC Inference

$0.002 / call

after plan limit

PCG Generation

$0.05 / mesh

after plan limit

TTS Synthesis

$0.008 / second

of generated audio

All plans include a 14-day free trial. No overages charged without explicit opt-in. Prices in USD.

Active — Early Access

Neural NPC Engine

Dynamic NPC dialogue powered by large language models. NPCs maintain emotional states, track conversation history, and respond in the player's language. The demo on the homepage is powered by this engine.

NPC_Memory_Stream.log

[08:42:11] EVENT: Player saved NPC from bandits.

[08:42:12] ROUTE: Model=Gemini Flash | Latency=180ms

[08:42:13] RAG: Vector search query 45ms

... 4 hours later ...

[12:15:02] EVENT: Player approaches.

[12:15:03] INFERENCE: "I owe you my life. I won't forget what you did."

[12:15:04] TTS_STREAM: Voice stream initiated (JP) E2E=540ms

Personality System

Each NPC has a defined personality, backstory, and behavioral rules in its system prompt. The LLM generates contextual, in-character responses that match the NPC's voice.

Emotion Tracking

NPCs output sentiment tags (Hostile, Suspicious, Neutral, Helpful, etc.) that reflect their emotional state based on the conversation. This drives UI feedback and game logic.

Multi-Language

NPCs automatically detect the player's language and respond in the same language. Works with any language the underlying LLM supports.

Current Capabilities

Live

NPC Dialogue Engine

Multi

Language Detection

Emotion

State Tracking

Voice synthesis (TTS/STT) and vector RAG memory are planned features — not yet available.

Use Cases

Open-World Quest Givers

NPCs dynamically generate quests based on the player's history, reputation, and current world state. No two playthroughs produce the same quest lines.

Player: "Any work around here?"
NPC: "After what you did at the mine... I trust you. There's something bigger. Meet me at dusk."

Intelligent Merchants

Shopkeepers who remember your purchases, haggle dynamically based on relationship, and gossip about events you've caused in the world.

Player: "Got any potions?"
NPC: "For you? Discount. You cleared those wolves from my supply route last week."

Companion Characters

Party members that form real relationships with the player over time, reacting emotionally to decisions and recalling shared history across dozens of hours.

Companion: "Remember that village we saved? I still think about it. Made me believe we're doing the right thing."

Adaptive Antagonists

Villains who taunt you based on your actual failures, adapt their strategies to your playstyle, and deliver monologues that reference your specific choices.

Boss: "You failed to save the bridge. You failed to save your friend. And now you'll fail here too."

Quick Integration

Unity C# Example

async Task<string> TalkToNPC(string npcId, string playerMsg) {
    var request = new NPCChatRequest {
        NpcId = npcId,
        PlayerId = GameSession.CurrentPlayer.Id,
        Message = playerMsg,
        ContextVars = new {
            time_of_day = WorldClock.TimeOfDay,
            player_health = Player.Health,
            weapon_drawn = Player.IsWeaponDrawn
        }
    };

    var response = await VantageClient.NPC.ChatAsync(request);

    // Apply emotion to animation system
    AnimController.SetEmotion(response.EmotionState);

    // Stream audio to game
    AudioManager.PlayStream(response.AudioUrl);

    return response.Dialogue;
}

Research — Not Yet Available

Procedural Gen Engine

Concept: generate 3D assets and level layouts from text prompts. This is early-stage research — the features described below represent our vision, not current capabilities.

Research stage: None of the features on this page are available for use. We're sharing our research direction for transparency.

1. Text Prompt

"A rusted cyberpunk vending machine, 4k textures"

2. Vantage PCG Cloud

Diffusion to 3D Mesh Generation (~15s processing)

3. Engine Import

Auto-imported as .FBX with linked Albedo/Normal maps

Asset Generator

Generate props, foliage, and clutter items directly inside your editor. Great for filling out environments quickly without searching asset stores.

Infinite Level Blueprint

Define rulesets (e.g., "Sci-fi corridor, claustrophobic, 5 rooms") and let the engine construct logical, playable map layouts dynamically at runtime.

Supported Output Formats

.FBX

3D Meshes

.GLTF

Web / Universal

.PNG

PBR Textures (4K)

.USD

Scene Composition

Prompt Examples & Results

"Crystal sword with glowing blue runes, fantasy RPG style"

Vertices: 2,847 Gen Time: 12s

"Ancient oak tree, 4 seasons LOD variants, game-ready"

LOD Levels: 4 Gen Time: 18s

"Underground dungeon, 8 rooms, traps, treasure room at end"

Rooms: 8 Gen Time: 6s

How It Works

1

Text-to-Shape Diffusion

Your text prompt is encoded into a latent space using a fine-tuned CLIP model. A 3D-aware diffusion model generates a signed distance field (SDF) representing the shape.

2

Mesh Extraction & Optimization

Marching cubes extract a polygon mesh from the SDF. Automatic retopology reduces polygon count while preserving visual detail. LOD variants are generated.

3

PBR Material Synthesis

A separate texture diffusion model generates Albedo, Normal, Roughness, and Metallic maps at 4K resolution, UV-mapped to the generated mesh.

4

Engine Delivery

The asset is packaged in your chosen format (.FBX/.GLTF/.USD) and delivered via our SDK directly into your editor's asset browser. No manual import needed.

Research — Not Yet Available

Anti-Cheat Sentinel

Concept: analyze player telemetry to detect suspicious behavior patterns. This is early-stage research — the features described below represent our vision, not current capabilities.

Research stage: This module is not available for use. We're sharing our research direction for transparency.

Security Dashboard

Live Monitoring

User_921A (Aimbot Suspected)

Crosshair snap speed exceeds human limit

Flagged

User_BB42 (Macro Script)

Input interval consistency warning

Review

Behavioral Telemetry

We don't scan local files. We look at *how* the player plays, analyzing input patterns that software cannot easily disguise.

Server-Side Validation

Sentinel telemetry processing runs entirely on your dedicated servers or our Edge Cloud, reducing client-side vulnerabilities.

Statistical Anomaly Detection

We compare player metrics against population baselines. Headshot ratios, reaction times, and movement patterns that deviate beyond 3 sigma trigger investigation flags.

Replay Analysis

Flagged sessions are automatically recorded. Your moderation team can review input-frame data, crosshair trajectories, and timing histograms before taking action.

What Sentinel Detects

Aim Assistance

Snap-to-target speed beyond human capability
Unnatural tracking smoothness on moving targets
FOV-limited awareness (seeing through walls)

Input Automation

Identical input intervals (macro detection)
Inhuman APM (actions per minute) sustained
Frame-perfect ability combos over time

Movement Exploits

Speed values exceeding server physics limits
Teleportation / position desync patterns
Noclip detection via collision map analysis

Integration Steps

1

Add SDK

Drop our lightweight agent into your game server (12KB footprint)

2

Stream Telemetry

Player input events streamed to Sentinel Edge in real-time

3

Get Verdicts

Receive confidence scores via webhook or dashboard alerts

4

Take Action

Auto-ban, shadow-ban, or escalate to human moderators

System Architecture

The NPC demo on this page connects to a live inference backend. Here's a simplified view of how the system works.

Game Client / Browser

Web, Unity, Unreal, Godot

REST API JSON Messages

HTTPS REST API

API LAYER

Express API Server

Node.js + SQLite

Session Auth Rate Limiting

API Proxy Sanitized

LLM Provider

Cloud Inference API

NPC Dialogue System Prompts

How the Demo Works

Request Flow

01 Player sends a message in the chat input
02 Message is POSTed to /api/inference
03 Server proxies to LLM provider with NPC system prompt
04 Response is sanitized (provider details removed)
05 Sentiment tag extracted, NPC reply displayed

Tech Stack

01 Node.js + Express API server
02 SQLite for user auth and sessions
03 Server-side API proxy (hides provider details)
04 Response sanitization layer
05 PM2 for process management

Documentation

Quick Start Guide

Learn how to initialize the Vantage SDK and make your first cognitive NPC request in under 5 minutes.

1. Installation

Install the core package via npm (for web/Node.js testing) or download the specific game engine SDK from your portal.

npm install @vantage-labs/core-sdk

2. Initialization (C# Unity Example)

Initialize the client with your studio API key. Make sure to keep this key secure and use environment variables in production.

using VantageLabs.Core;
using UnityEngine;

public class GameManager : MonoBehaviour {
    void Start() {
        // Initialize with your portal API Key
        VantageClient.Init("vk_live_8f92a1...");

        Debug.Log("Vantage Engine Connected!");
    }
}

3. Your First NPC Conversation

Send a message to an NPC and handle the response. The response includes dialogue text, audio URL, emotional state changes, and optional animation triggers.

var response = await VantageClient.NPC.ChatAsync(
    npcId: "npc_blacksmith_01",
    playerId: Player.Id,
    message: "Can you repair my sword?",
    contextVars: new { player_gold = 150, reputation = "friendly" }
);

// response.Dialogue = "Aye, I can fix that. 50 gold and it'll be sharper than new."
// response.EmotionState = "helpful"
// response.AudioUrl = "https://cdn.vantage.dev/tts/abc123.mp3"
// response.ActionTrigger = "anim_inspect_item"

4. Handling WebSocket Events

For real-time streaming (recommended for dialogue), subscribe to WebSocket events instead of using REST polling.

VantageClient.OnDialogueChunk += (chunk) => {
    UIManager.AppendDialogueText(chunk.Text);
};

VantageClient.OnAudioReady += (audio) => {
    AudioSource.PlayStream(audio.StreamUrl);
};

VantageClient.OnEmotionShift += (emotion) => {
    NPCAnimator.SetTrigger(emotion.AnimTrigger);
};

5. Configuration Options

Customize behavior per-NPC using the configuration object:

Parameter	Type	Description
temperature	float	Creativity of responses (0.1-1.5)
max_tokens	int	Max response length (default 256)
voice_id	string	TTS voice model identifier
memory_enabled	bool	Enable episodic memory for this NPC
personality	string	System prompt defining NPC character

Unity Integration

Step-by-step guide to integrating the Vantage SDK into your Unity project.

1. Import the Package

Download the Unity package from your developer portal or install via the Unity Package Manager using our git URL.

Window > Package Manager > + > Add from git URL:
https://github.com/vantage-digital-labs/unity-sdk.git

2. Create VantageManager

Add the VantageManager component to an empty GameObject in your scene. This handles connection lifecycle and token refresh.

using VantageLabs.Unity;

public class MyGame : MonoBehaviour {
    [SerializeField] string apiKey;

    void Awake() {
        VantageManager.Instance.Initialize(apiKey);
        VantageManager.Instance.OnConnected += () => {
            Debug.Log("Vantage WebSocket ready");
        };
    }
}

3. Attach NPCBrain Component

Add an NPCBrain to any character that needs cognitive dialogue. Configure personality and voice in the Inspector.

var brain = npcObject.GetComponent<NPCBrain>();
brain.SetPersonality("A grumpy blacksmith who secretly has a heart of gold");
brain.SetVoice("voice_gruff_male_01");

// Player initiates dialogue
var reply = await brain.Say("Can you fix my armor?");
dialogueUI.ShowText(reply.Text);
audioSource.PlayStream(reply.AudioClip);

4. Platform Requirements

Requirement	Minimum
Unity Version	2021.3 LTS+
.NET	.NET Standard 2.1
Platforms	Windows, macOS, Linux, iOS, Android
Dependencies	Newtonsoft JSON (auto-included)

Unreal Engine Plugin

Native C++ plugin for Unreal Engine 5.2+ with Blueprint support.

1. Plugin Installation

Clone the plugin into your project's Plugins directory:

git clone https://github.com/vantage-digital-labs/unreal-plugin.git Plugins/VantageLabs

2. Enable Plugin

In your .uproject file, add the plugin dependency. Then restart the editor.

{
  "Plugins": [
    { "Name": "VantageLabs", "Enabled": true }
  ]
}

3. C++ Usage

#include "VantageNPC.h"

void AMyNPC::TalkTo(FString PlayerMessage) {
    FVantageRequest Req;
    Req.NpcId = NpcId;
    Req.Message = PlayerMessage;

    UVantageSubsystem* Vantage = GetGameInstance()->GetSubsystem<UVantageSubsystem>();
    Vantage->SendChat(Req, FOnChatResponse::CreateLambda([this](FVantageResponse Res) {
        DialogueWidget->SetText(Res.Dialogue);
        AudioComp->PlayStreamFromURL(Res.AudioUrl);
    }));
}

4. Blueprint Support

All core functions are exposed to Blueprints. Drag the "Vantage NPC Chat" node into any event graph and connect your NPC ID and player input pins.

Authentication

How to authenticate your game client with the Vantage API securely.

API Key Types

Key Prefix	Environment	Usage
vk_test_	Sandbox	Development & testing (no billing)
vk_live_	Production	Live game builds (metered billing)

Request Authentication

Include your API key in the Authorization header of every request:

Authorization: Bearer vk_live_8f92a1b3c4d5...

Token Refresh

API keys do not expire, but can be rotated from your developer portal. For WebSocket connections, a session token is issued on handshake and auto-refreshes every 30 minutes.

Security Best Practices

Never embed API keys in client-side code shipped to players
Use environment variables or a secure vault in your game server
Rotate keys immediately if compromised
Use vk_test_ keys during development

WebSocket Events

Real-time streaming via persistent WebSocket connections for minimal latency dialogue.

Connection Lifecycle

// Connect to regional endpoint
const ws = new WebSocket("wss://rt.vantage.dev/v2/stream");

ws.onopen = () => {
    ws.send(JSON.stringify({
        type: "auth",
        token: "vk_live_..."
    }));
};

ws.onmessage = (event) => {
    const data = JSON.parse(event.data);
    switch(data.type) {
        case "chunk": handleDialogueChunk(data); break;
        case "audio": handleAudioReady(data); break;
        case "emotion": handleEmotionShift(data); break;
        case "error": handleError(data); break;
    }
};

Event Types

Event	Description	Payload
chunk	Partial dialogue token	{text, index}
audio	TTS audio URL ready	{url, duration_ms}
emotion	NPC emotion state change	{state, intensity}
action	Animation trigger	{trigger_name}
error	Pipeline error	{code, message}

API Reference

NPC ENGINE v2

WORLD GEN v1

POST/v1/world/mesh

Back to Docs

POST

/v2/npc/chat

Send a player message to an NPC entity. The engine generates a cognitive response including localized dialogue text, synthesized audio URLs, emotional state shifts, and engine action triggers.

Parameters

npc_id Required

String. The unique identifier for the NPC instance.
message Required

String. The text input from the player.
context_vars Optional

Object. Real-time game state variables to influence AI decision making (e.g. time of day, player health).

Request Body Example

{
  "npc_id": "npc_cyber_bartender_01",
  "player_id": "usr_98213",
  "message": "I'm looking for the black market.",
  "context_vars": {
    "player_weapon_drawn": false,
    "time_of_day": "night"
  }
}

Response 200 OK

{
  "dialogue": "Take a left at the neon alley...",
  "audio_url": "https://cdn.vantage.dev/tts/x.mp3",
  "emotion_state": "nervous",
  "action_trigger": "anim_look_around",
  "affinity_change": -1,
  "latency_ms": 487
}

Authentication

All API requests require a Bearer token in the Authorization header. Obtain your API key from the developer portal after registration.

Authorization: Bearer vk_live_8f92a1b3c4d5e6f7...

Rate Limits

Plan	Requests/min	Monthly Limit	Burst
Sandbox	20	5,000	5
Pro	300	100,000	50
Enterprise	Custom	Unlimited	Custom

Error Codes

400

Bad Request — Missing required fields or invalid parameter types.

401

Unauthorized — Invalid or expired API key.

429

Rate Limited — You've exceeded your plan's rate limit. Retry after the indicated delay.

500

Internal Error — Inference pipeline failure. Automatic retry recommended.

Official SDKs

Unity SDK

C# | v2.4.1

Unreal Plugin

C++ | v2.4.0

Node.js SDK

TypeScript | v2.4.0

GET

/v2/npc/memory

Retrieve the episodic memory entries stored for a specific NPC-player pair. Returns the most recent interactions and their computed emotional impact.

Query Parameters

npc_id Required

The NPC whose memory to retrieve.
player_id Required

The player whose interactions to filter by.
limit Optional

Max entries to return (default 20, max 100).

Response 200 OK

{
  "npc_id": "npc_blacksmith_01",
  "player_id": "usr_98213",
  "memory_entries": [
    {
      "timestamp": "2026-05-14T08:42:11Z",
      "event": "Player repaired sword",
      "emotion_impact": { "trust": +15, "joy": +5 },
      "relevance_score": 0.87
    }
  ],
  "current_affinity": 72,
  "emotion_state": "trusting"
}

POST

/v1/world/mesh

Generate a 3D mesh asset from a text prompt. Returns download URLs for the generated mesh and PBR texture maps.

Request Body

{
  "prompt": "A rusted cyberpunk vending machine",
  "format": "fbx",
  "texture_resolution": 2048,
  "lod_levels": 3,
  "style_preset": "game_ready"
}

Parameters

prompt Required

Text description of the asset to generate (max 500 chars).
format Optional

Output format: "fbx", "gltf", "usd" (default: "fbx").
texture_resolution Optional

Texture map resolution: 1024, 2048, 4096 (default: 2048).

Response 202 Accepted

{
  "job_id": "mesh_job_a7b3c9",
  "status": "processing",
  "estimated_time_s": 15,
  "poll_url": "/v1/world/mesh/mesh_job_a7b3c9/status"
}

Mesh generation is asynchronous. Poll the status URL or subscribe to webhook notifications.

Privacy Policy

Last Updated: May 1, 2026 (v3.2 — updated service regions)

Previous revision: November 8, 2025 (v3.1)

1. Information We Collect

Account data: When you register, we collect your email address, studio name, and a hashed password. We do not require real names or billing information for free-tier accounts.

API usage data: We log request metadata (timestamp, endpoint, response time, HTTP status) for monitoring and abuse prevention. We do not log the content of NPC dialogue or player messages passed through our inference API unless you explicitly enable the episodic memory feature.

Website demo: The NPC demo on this page sends your chat messages to our inference API for response generation. Messages are processed in real-time and are not stored after the response is returned.

Voice data: Audio processed through our TTS/STT pipeline is ephemeral — it is processed in memory and never written to disk. No voice recordings are retained after the inference response is returned.

2. How We Use Information

Your information is used solely to: authenticate API requests and enforce rate limits; send product updates and security notices (if you subscribe); analyze aggregate usage patterns (e.g., total requests per region) to optimize infrastructure scaling. We do not profile individual users or sell data to advertisers.

3. Data Storage & Security

All data is encrypted in transit (TLS). Passwords are hashed using bcrypt. The database uses SQLite stored on the server filesystem. NPC dialogue data processed through the demo is ephemeral — messages are not stored after the response is returned.

4. Sub-Processors

We use the following sub-processors:

LLM Inference Providers — Third-party model providers for NPC dialogue generation (website demo only)

We do not sell or share your data with any other third parties.

5. Your Rights

You may request a copy of all data we hold about you, or request permanent deletion, by emailing store@vantage-digital.online. We will fulfill data requests within 30 days. For studios using our API, deletion includes removal of all API keys, session tokens, and stored NPC memory data.

6. Local Storage

This website uses browser localStorage to maintain your login session in the developer portal. We do not use cookies, tracking pixels, analytics scripts (e.g., Google Analytics), or advertising networks.

7. Contact

Vantage Digital Labs
Email: store@vantage-digital.online
GitHub: github.com/vantage-digital-labs

Terms of Service

Effective Date: May 1, 2026 (v4.0 — updated service tiers)

Previous revision: January 20, 2026 (v3.8)

1. Acceptance of Terms

By creating an account or making API requests to Vantage Digital Labs services, you agree to these Terms. If you are using the service on behalf of a studio or company, you represent that you have authority to bind that entity.

2. Service Description

Vantage provides AI tooling for game teams, currently including:

Cognitive NPC Engine — Dialogue generation with emotional state tracking and multi-language support (active, early access)
NPC Demo — Live demo on this website powered by our inference API proxy

Additional modules (Procedural Gen, Anti-Cheat, Voice AI) are in research stage and not yet available.

3. API Usage & Rate Limits

Each pricing tier has defined rate limits published on our pricing page. Exceeding limits results in HTTP 429 throttling, not billing charges. The free tier is limited to 5,000 NPC inference calls per month and is intended for evaluation and prototyping — using it for production game traffic may result in account suspension. Enterprise clients negotiate custom limits in their Master Service Agreement.

4. Data Processing

NPC dialogue content is processed in-memory and not persisted. User account data (email, hashed password) is stored in a local SQLite database. We do not currently offer data residency options or regional deployment.

5. Intellectual Property

Content generated through our APIs (NPC dialogue text) is owned by you. Vantage retains no rights to generated outputs. You may use generated content in commercial games without attribution. Our inference infrastructure, system prompts, and code remain Vantage's intellectual property.

6. Service Level Agreement

All tiers are provided on a best-effort basis. We do not currently offer uptime SLAs or downtime credits. We aim to maintain availability and will communicate planned maintenance in advance.

7. Limitation of Liability

Vantage is not liable for indirect, incidental, or consequential damages (including lost revenue from game downtime). Total liability is capped at fees paid in the 12 months preceding the claim. This limitation applies to the fullest extent permitted by applicable law.

8. Termination

Either party may terminate with 30 days written notice. On termination: API keys are revoked immediately, stored NPC memory data is deleted within 30 days, and any prepaid fees for unused months are refunded on a pro-rata basis. Enterprise contracts may have separate termination clauses in the MSA.

9. Governing Law & Disputes

These Terms are governed by the laws of the jurisdiction where Vantage Digital Labs operates. Disputes will be resolved through good-faith negotiation first. If unresolved, disputes will be settled through binding arbitration. Either party may seek injunctive relief in any court of competent jurisdiction for IP infringement.

Security

Last Updated: May 22, 2026

Honest disclosure: We are a small team in early access. We have not completed external security audits (SOC 2, ISO 27001, or penetration testing). The practices below describe what we actually do today.

1. Data in Transit

All connections to our API and website use TLS 1.2+ encryption. HTTP requests are redirected to HTTPS. API keys are transmitted only in Authorization headers — never in URLs or query parameters.

2. Data at Rest

User passwords are hashed using bcrypt with a cost factor of 10. Session tokens are UUIDs stored in SQLite. The database file is stored on the server filesystem — we do not use external database services at this stage. We do not store NPC conversation content beyond the active session.

3. Authentication

User accounts use email + password authentication. Session tokens are generated on login/register and validated on each authenticated request. Tokens can be invalidated via the logout endpoint. We do not implement OAuth, SSO, or MFA at this time.

4. API Security

The inference API proxy hides third-party provider details from the frontend. Provider API keys are stored server-side only. Rate limiting is handled at the application level per pricing tier.

5. What We Don't Do (Yet)

No external security audit or penetration testing
No SOC 2 or ISO 27001 certification
No bug bounty program
No multi-factor authentication (MFA)
No WAF or DDoS protection beyond what the hosting provider offers

As we grow, we plan to implement these. If you have security concerns or find a vulnerability, please contact us at store@vantage-digital.online.

6. Infrastructure

The website and API run on a single server using Node.js with Express. The SQLite database provides local persistence. We use PM2 for process management. The site is served behind a reverse proxy with HTTPS termination.

7. Reporting Vulnerabilities

If you discover a security vulnerability, please report it responsibly to store@vantage-digital.online. We aim to acknowledge reports within 48 hours and provide a fix timeline based on severity. We appreciate responsible disclosure and will credit reporters (with permission).

AI Tooling for Game Teams

What We Build

AI Tooling for Game Teams.

Contact

About Us

What We're Building

AI Modules for Game Teams

Cognitive NPCs

Procedural World Gen

Anti-Cheat Analysis

How It Works

Inference Proxy

NPC System Prompts

Language Detection

Experience Cognitive NPCs

Operator-7

Terminal Connection_

How This Demo Works

Try These Example Inputs

Transparent Pricing

Sandbox

Pro

Enterprise

Usage-Based Overage Pricing

Stay in the Loop

Neural NPC Engine

NPC_Memory_Stream.log

Personality System

Emotion Tracking

Multi-Language

Current Capabilities

Use Cases

Open-World Quest Givers

Intelligent Merchants

Companion Characters

Adaptive Antagonists

Quick Integration

Procedural Gen Engine

1. Text Prompt

2. Vantage PCG Cloud

3. Engine Import

Asset Generator

Infinite Level Blueprint

Supported Output Formats

Prompt Examples & Results

How It Works

Text-to-Shape Diffusion

Mesh Extraction & Optimization

PBR Material Synthesis

Engine Delivery

Anti-Cheat Sentinel

Security Dashboard

User_921A (Aimbot Suspected)

User_BB42 (Macro Script)

Behavioral Telemetry

Server-Side Validation

Statistical Anomaly Detection

Replay Analysis

What Sentinel Detects

Aim Assistance

Input Automation

Movement Exploits

Integration Steps

Add SDK

Stream Telemetry

Get Verdicts

Take Action

System Architecture

Game Client / Browser

Express API Server

LLM Provider

How the Demo Works

Request Flow

Tech Stack

Documentation

Quick Start Guide

1. Installation

2. Initialization (C# Unity Example)

3. Your First NPC Conversation

4. Handling WebSocket Events

AI Tooling for
Game Teams