WRW Agentic Engine
A Domain-Agnostic AI Swarm on GCP

Author: William Ruel Wilmoth, Architect | Version: 4.0 | Date: May 2026 | Status: Patent Pending | Contact: bill@wrw-systems.dev

Abstract

This whitepaper details the design and deployment of the WRW Agentic Engine—a fully autonomous, 12-node AI swarm running natively on Google Cloud Platform (GCP). Built by a non-technical architect using Gemini as a co-pilot, the Engine translates the physical logistics of a factory floor into a reliable digital workflow. By coupling strict constitutional governance, an asynchronous ledger-based state manager, and isolated Google Assured OSS pipelines, the Engine achieves high-fidelity, hallucination-resistant reasoning through multi-agent debate.

1. Introduction

The Factory Floor as a Cognitive Model

Modern multi-agent AI systems frequently fail at scale due to context degradation, unchecked hallucination, and unstructured dependency loops. The Engine solves these crippling issues by abandoning traditional software paradigms in favor of a philosophically pure digital assembly line.

Drawing on blue-collar industrial experience, the Engine replaces unstructured agent chat with an asynchronous, physical ledger. Agents do not "talk" to each other. They read raw data from previous waves on a ledger and write structured outputs for the next wave. Through a human-led "Write-Read-Write" methodology, the Architect utilized Google's Gemini models to rapidly generate the Python and GCP configurations necessary to bring this disciplined, physical logic into the cloud.

2. Infrastructure

The Pivot to Google Cloud

Initial attempts to build the Engine using competing stacks resulted in a bottleneck of dependency friction. Transitioning to the Google Cloud ecosystem unlocked the velocity needed to deploy the entire 12-node architecture from zero to production in just 30 days.

• Firestore: The rapid, transactional state manager. It handles the asynchronous read/writes required by the execution loops without bottlenecking parallel agent tasks.

• BigQuery: The deep-storage analytics pipeline. It aggregates telemetry and Engine Health Metrics, allowing for SQL-based statistical analysis over thousands of cycles.

• Vertex AI API: Powers the core intelligence, facilitating the real-time execution of Gemini models within strict environmental constraints.

3. Core Architecture

The 12-Node Swarm

The Engine relies on a rigid hierarchical structure composed of exactly 12 specialized nodes, segmented into functional 3-agent sleeves and a 3-agent C-Suite. This modularity makes the system entirely domain-agnostic. While the current instance is optimized for synthetic algorithmic trading, the functional sleeves can be seamlessly repurposed for any data-heavy vertical.

• Wave 1: The Leads (ML-Research-1, Code-Writer-1, Execution-Trader-1): Drive initial logic drafting and execution planning. They parse raw data vaults and translate requirements into executable formats.  

• Wave 2: The Critics (ML-Research-2, Code-Writer-2, Execution-Trader-2): The internal immune system. They hunt for fatal flaws, security vulnerabilities, and overfit logic by engaging the Leads in a strict 3-round debate protocol.  

• Wave 3: Supervisors (ML-Research-Supervisor, Code-Writer-Supervisor, Trader-Supervisor): The adjudication layer. They resolve deadlocks and enforce strict risk constraints before advancing logic to the C-Suite. Supervisors actively cross-pollinate Engine intelligence.  

• Wave 4: The C-Suite (Manager-Agent, Operations-Agent, Constitution-Agent): Handles macro-strategy, system health monitoring, and Constitutional compliance. They manage physical capital overrides, perform unmediated debugging audits, and execute Epoch Consolidations.

4. Safety & Security

The Constitution

High-stakes autonomous deployment requires absolute trust in the system's ability to resist hallucination and insecure code execution. The Engine’s Constitution enforces this through several critical, unyielding operational laws, including:

The 3-Key Lock & Google Assured OSS Provenance: To prevent supply chain attacks and hallucinated dependencies, the Code-Writer Sleeve is prohibited from accessing the open internet. Unanimous consent is required to execute a Sandbox environment:

• Code-Writer-1 drafts logic strictly utilizing officially maintained Google SDKs and Google Assured OSS packages.

• Code-Writer-2 audits the script specifically for unauthorized external dependencies.

• Code-Writer-Supervisor exercises final veto power. If the 3-Key Lock fails, the supervisor orders a sterile rebuild in the ledger.

The Ideological Swap Protocol: When a deadlock occurs between a Lead and a Critic, standard summarization is bypassed. To prevent ideological anchoring, both agents are forced to argue the opponent's strongest empirical point in one paragraph before the Supervisor rules. This mechanically breaks logic loops.

The 5-Flag Diet Telemetry: To optimize cognitive bandwidth, agents communicate outcomes to the physical pipeline via strict algorithmic tags rather than natural language summaries.

Example Output Telemetry: [FLAG: CRITICAL CONSENSUS ACHIEVED] - Sandbox logic verified and locked.  

5. Conclusion

Domain-Agnostic Reason

The WRW Agentic Engine proves that complex, 12-node orchestration does not require sprawling codebases or massive engineering teams. By applying the physical discipline of a factory assembly line to GCP's robust architecture—and utilizing Gemini as an AI co-pilot—one non-technical architect designed and deployed a highly secure, hallucination-resistant Reason Engine in 30 days. It stands as a production-ready framework for mission-critical AI deployment across any vertical that demands strict execution boundaries.