01.0

Neural Architecture Overview

Workiza represents a paradigm shift from **User-Initiated** AI (Chatbots) to **Event-Driven** AI (Autonomous Agents). The architecture is designed to function as a "shadow workforce" that operates asynchronously alongside your human teams.

The Trigger Layer

Agents listen for JSON payloads from webhooks (Zapier, Make, Segment) or direct API calls. No human interface is required to start a job.

The Execution Layer

Once triggered, the agent spins up a dedicated container, accesses MindFiles™ for context, performs the task, and terminates.

02.0

Agent Chassis Specifications

The Fabricator Console allows for the deployment of three distinct chassis types, each optimized for specific cognitive loads and output formats.

Lead Qualifier (L4_EVENT)

  • Input: Raw lead data (Name, Email, Company).
  • Process: Cross-references LinkedIn API, Crunchbase, and Company News.
  • Output: 0-100 Lead Score + Personalized Email Draft.
  • Cost: $0.45 per execution.

Deep Researcher (L5_SWARM)

  • Input: Topic, Ticker Symbol, or Competitor URL.
  • Process: Recursive web crawling (Depth: 3 layers). Sentiment analysis of news.
  • Output: 5-10 Page PDF Intelligence Dossier with citations.
  • Cost: $2.50 per execution.
03.0

Integration Protocols

Developer Access Restricted

The Neural API is currently in Private Beta for enterprise partners. Public access to programmable endpoints (v1/trigger) is scheduled for Q4 2026. To request early access keys, please contact support.

Once authorized, you may trigger agents programmatically via a POST request to your specific Agent Endpoint.

Future V1 Spec: JSON Payload

POST https://api.workiza.ai/v1/trigger
Content-Type: application/json
Authorization: Bearer wk_live_99821...

{
  "agent_id": "agn_71pm8bmh4",
  "context": {
    "target_company": "nvidia.com",
    "priority": "high",
    "directives": [
      "Analyze Q3 earnings",
      "Check for insider trading signals"
    ]
  }
}
04.0

MindFiles™ Security

MindFiles™ is the proprietary shared memory layer of Workiza. Unlike standard LLM context windows which reset, MindFiles™ persists key corporate data in a vector database encrypted with **AES-256**.

SOC2 Type II Compliance

All data processing occurs within isolated containers. Data is encrypted in transit (TLS 1.3) and at rest. We do not train our foundational models on customer MindFiles™ data without explicit written consent.