Vouchstone Documentation

What is Vouchstone?

Vouchstone is an enterprise-grade AI agent platform that enables organizations to build, deploy, and manage intelligent agents with persistent long-term memory. Unlike traditional AI systems that forget context between interactions, Vouchstone agents remember, learn, and evolve over time.

Our unique Control Plane + Data Plane architecture ensures that your sensitive data never leaves your infrastructure while providing a seamless management experience through our cloud-hosted control plane.

Key Features

Intelligent Agents

Create AI agents that can understand context, remember past interactions, and make intelligent decisions. Each agent can be customized with specific capabilities, personality traits, and access to different memory stores.

Memory Systems

Vouchstone provides a biologically-inspired 5-layer memory stack, each layer analogous to a region of the human brain:

• Working Memory (Redis) — In-context state for the current turn, bounded by context window, resets at session end. Analogous to prefrontal cortex.
• Episodic Memory (PostgreSQL + Redis) — Append-only traces of what happened across sessions, retrieved at planning time. Analogous to hippocampus.
• Semantic Memory (ChromaDB) — Entity store of typed attributes, built by extraction jobs from episodes. Analogous to temporal lobe.
• Procedural Memory (Neo4j) — Versioned skill registry of procedures, built by reflection passes over episodes. Analogous to basal ganglia + cerebellum.
• Meta-Memory (Control plane) — Governs the other four layers: retention policies, decay, compression, dedup, forgetting. Analogous to anterior prefrontal cortex.

Workflow Automation

Build complex automated workflows that orchestrate multiple agents, integrate with external systems, and handle conditional logic. Our visual workflow builder makes it easy to create sophisticated automation without code.

Multi-Agent Collectives

Group agents into collectives for collaborative problem-solving. Collectives can use different orchestration strategies including hierarchical, democratic voting, and custom coordination patterns.

Architecture

Vouchstone uses a unique split architecture that separates the management layer from your data:

Getting Started

Follow these steps to get up and running with Vouchstone:

Quick Examples

Python SDK

from vouchstone_sdk import VouchstoneClient

# Initialize the client
client = VouchstoneClient(api_key="your-api-key")

# Create an agent
agent = client.agents.create(
    name="Customer Support Agent",
    type="support",
    memory_types=["semantic", "episodic"],
    config={
        "personality": "helpful and professional",
        "response_style": "concise"
    }
)

# Send a message and get a response
response = agent.chat("How do I reset my password?")
print(response.content)

TypeScript SDK

import { VouchstoneClient } from '@vouchstone/sdk';

// Initialize the client
const client = new VouchstoneClient({ apiKey: 'your-api-key' });

// Create an agent
const agent = await client.agents.create({
  name: 'Data Analyst',
  type: 'analyst',
  memoryTypes: ['semantic', 'procedural'],
  config: {
    capabilities: ['data_analysis', 'visualization'],
  },
});

// Query the agent
const response = await agent.chat('Analyze sales trends for Q4');
console.log(response.content);