GotiHub AGL - Governance-First AI Workflows

Governance-first AI workflows using local Gemma reasoning and Midnight Zero-Knowledge verification.

Comment

Inspiration

GotiHub AGL began from a real institutional problem.

Many organizations still rely on spreadsheets, manual approvals, phone calls, and fragmented verification processes for sensitive workflows like alumni validation, compliance reviews, and governance approvals.

At the same time, AI adoption is accelerating rapidly — but most institutions remain uncomfortable sending sensitive records to external AI providers.

That created a trust gap.

I wanted to explore a different direction:

What if AI reasoning could stay local, governance could remain human-controlled, and verification could become cryptographically auditable?

That question became the foundation of GotiHub AGL.

What it does

GotiHub AGL is a governance-first AI workflow platform designed for institutional approvals, compliance operations, and verification systems.

The platform combines:

  • Local AI reasoning using Gemma via Ollama
  • Human governance escalation workflows
  • Risk-aware approval orchestration
  • Midnight Zero-Knowledge verification
  • Immutable governance audit trails

Instead of replacing governance with AI, the system uses AI to assist institutional decision-making while keeping humans in control.

GotiHub AGL — Governance-First AI Workflows

Inspiration

GotiHub AGL began from a real institutional problem.

Many organizations still rely on spreadsheets, manual approvals, phone calls, and fragmented verification processes for sensitive workflows like alumni validation, compliance reviews, and governance approvals. At the same time, AI adoption is accelerating rapidly — but most institutions remain uncomfortable sending sensitive, private records to external cloud AI providers.

That created a massive trust gap. I wanted to explore a completely different direction:

What if AI reasoning could stay local, governance could remain human-controlled, and verification could become cryptographically auditable?

That core question became the foundation of GotiHub AGL.

What it does

GotiHub AGL is a governance-first AI workflow platform designed for institutional approvals, compliance operations, and verification systems. The platform combines:

  • Local AI Reasoning: Secure, local intelligence using Gemma via Ollama.
  • Human Governance Escalation: Advanced multi-stage workflows that flag anomalies automatically.
  • Risk-Aware Orchestration: Dynamically routing workflows based on AI confidence scores and governance thresholds.
  • Midnight Zero-Knowledge Verification: Decentralized proof generation to maintain institutional confidentiality.
  • Immutable Governance Audit Trails: Permanent, verifiable logs that bridge cryptographic precision with administrative visibility.

Instead of replacing human governance with autonomous AI, the system uses local intelligence to assist institutional decision-making while keeping humans firmly in control.

Governance Workflow

       [ Request Submission ]
                  │
                 ▼
    [ Local AI Governance Review ]
                  │
                 ▼
         [ Risk Assessment ]
          /             \
    (Low Risk)       (High Risk)
        /                         \
     ▼                           ▼
[ Auto-Approved ]   [ Human Escalation ]
       \                   /
        ▼                 ▼
   [ Midnight ZK Proof Generation ]
                 │
                 ▼
     [ Immutable Audit Trail ]

The AI engine analyzes institutional records locally and generates:

  • governance reasoning
  • risk analysis
  • verification confidence
  • approval recommendations

High-risk workflows automatically escalate to human review.

How we built it

The project consists of three connected repositories working together.

┌────────────────────────────────────────────────────────┐
│              🏛️ GotiHub AGL (Core Platform)            │
│       [Laravel 13] [Filament Panels] [MySQL] [Nginx]   │
└───────────────┬────────────────────────┬───────────────┘
                │                        │
     (Internal Bus / IPC)          (Async Events / WS)
                │                        │
                ▼                        ▼
┌──────────────────────────────┐ ┌──────────────────────────────┐
│   🤖 Laravel AGL (Service)   │ │ ⚡ Midnight Bridge (Sidecar) │
│  [Local Gemma-2 via Ollama]  │ │   [Node.js / Bun Core API]   │
└──────────────────────────────┘ └──────────────────────────────┘

1. GotiHub AGL (Core Platform)

The main governance dashboard and workflow orchestration platform.

Built using:

  • Laravel 13
  • Filament
  • MySQL
  • Docker

  • Nginx

  • Backend Framework: Built using Laravel 13 for enterprise-grade service routing and model lifecycle management.

  • Administration Interface: Fully driven by Filament, utilizing custom page schemas, badge indicators, and dynamic infolists to render the real-time governance feeds.

  • Infrastructure: Wrapped inside specialized Alpine Docker containers running behind an Nginx reverse proxy.

2. Laravel AGL (Governance Service Layer)

A specialized, application-level service block responsible for handling our dynamic intelligence layer.

  • AI Orchestration: Integrated with Ollama via Laravel\Ai\Ai to pipeline data payloads into local models.
  • Deterministic Parsing: Uses precise regular expressions to extract tokenized values like DECISION: and RISK_SCORE: from raw model reasoning blocks.
  • Escalation Logic: Determines risk score weight thresholds, preventing corrupted payloads or mismatched input structures from executing auto-approvals.

3. Midnight Bridge

A standalone cryptographic verification service designed for privacy-preserving institutional workflows.

  • Runtime Environment: Powered by Node.js 22 (or Bun) for lightweight execution.
  • Security Isolation: Communicates with the main Laravel app over an isolated virtual bridge network, ensuring that sensitive private keys and circuit execution parameters never interact directly with web-facing entry points.
  • ZK-Proofs: Generates secure Zero-Knowledge proofs to protect alumni identity while providing public audit verification.

Challenges we ran into

The biggest challenge was not frontend development.

The hardest part was designing a system where:

  • AI assists governance
  • humans remain in control
  • workflows remain explainable
  • institutional privacy remains protected

We also faced major infrastructure challenges while integrating:

  • Laravel + Filament
  • local Gemma reasoning through Ollama
  • Docker orchestration
  • Midnight proof workflows
  • cross-container networking
  • AI runtime stability

Another challenge was ensuring the platform remained usable even when running fully local AI models on limited hardware resources.

Accomplishments that we're proud of

We are proud that GotiHub AGL successfully combines:

  • local AI institutional workflows
  • human approval escalation
  • risk-aware institutional verification
  • Midnight cryptographic proof integration
  • reusable Laravel governance architecture

Some highlights include:

  • Building a reusable Laravel governance package
  • Running local Gemma reasoning with Ollama
  • Creating governance-aware AI workflows
  • Integrating Zero-Knowledge proof concepts into institutional approvals
  • Designing a governance-first AI architecture instead of a purely automation-first approach

What we learned

This project fundamentally changed how we think about AI systems.

We realized the future is not simply:

“AI automation.”

The future is:

governed AI systems.

As AI becomes more capable, organizations will increasingly require:

  • governance-aware AI infrastructure
  • explainable workflows
  • cryptographic auditability
  • policy-aware reasoning
  • privacy-preserving verification systems

We also learned a tremendous amount about:

  • AI orchestration
  • Docker infrastructure
  • local LLM deployment
  • Laravel package architecture
  • governance workflow design

What's next for GotiHub AGL - Governance-First AI Workflows

Our long-term vision extends far beyond alumni verification.

Future roadmap includes:

  • multi-agent governance systems
  • cross-institution verification workflows
  • decentralized governance infrastructure
  • AI policy orchestration
  • privacy-preserving compliance networks
  • advanced Midnight proof integrations
  • federated institutional governance

We believe governance-aware AI infrastructure will become an essential software category as organizations adopt AI into critical workflows.

Because in high-trust environments:

AI can recommend. Governance decides.

Built With

Share this project:

Updates