For Technical Teams

Platform Architecture.

A technical deep-dive into the infrastructure, security, integrations, and SLAs behind MoeCloud's autonomous AI agent platform.

GCP-NativeServerlessSOC 2-Aligned140+ Functions
Architecture

Six Layers. Zero Servers to Manage.

Every component is fully managed, auto-scaling, and globally distributed on Google Cloud Platform.

01

AI Reasoning Engine

Anthropic Claude (Haiku/Sonnet) via shared callLLM abstraction. Prompt engineering with agent-specific personas, guardrails, and confidence thresholds.

llmClient.ts
02

Serverless Compute

Firebase Cloud Functions v2 (Gen 2) on Google Cloud Run. Node.js 22 runtime, auto-scaling, 900s max timeout for long-running pipelines.

140+ functions
03

Persistent State

Firestore for all operational data — tickets, leads, invoices, agent health, action logs. Real-time listeners for dashboard and agent coordination.

80+ collections
04

Object Storage

Google Cloud Storage for campaign assets, video renders, audio files, and document archives. Signed URLs for secure access.

GCS buckets
05

Voice & Telephony

Twilio for inbound/outbound calls with StatusCallback tracking. ElevenLabs for agent TTS voices. DTMF fallback IVR.

Twilio + ElevenLabs
06

Video Pipeline

HeyGen Avatar IV for talking-head renders. Submagic for post-production (captions, B-roll, effects). Upload-Post for multi-platform publishing.

HeyGen → Submagic → Upload-Post
Security

Enterprise-Grade Safety. Built In.

Every layer of the platform is designed with security, compliance, and auditability as first-class requirements.

Secrets Management

All API keys and credentials stored in GCP Secret Manager. No plaintext secrets in code or environment variables. Runtime-only access via defineSecret().

Auth & Access Control

Firebase Authentication with role-based permissions. Portal access gated by canAccessPage() middleware. Admin/viewer/agent role hierarchy.

Data Encryption

All data encrypted at rest (AES-256) and in transit (TLS 1.3). Firestore and GCS use Google-managed encryption keys by default.

Audit Trail

Every agent action written to action_log with agent_id, timestamp, description, and outcome. Full audit trail queryable by client, agent, or time range.

Human Approval Gates

Critical decisions require Moses approval via approval_queue. Confidence thresholds prevent autonomous action on low-certainty classifications.

Circuit Breakers

Pre-flight resource checks before expensive pipeline runs. Submagic credit exhaustion, ElevenLabs character limits, and HeyGen quota monitoring.

Integrations

Eight Core Integrations. One Platform.

Voice & SMS

Twilio

Inbound/outbound calls, SMS, StatusCallback webhooks

Voice AI

ElevenLabs

TTS generation for 5 agent voices (turbo v2.5)

Video AI

HeyGen

Avatar IV talking-photo video rendering

Post-Production

Submagic

AI captions, B-roll, zoom effects, music

Social Publishing

Upload-Post

TikTok, Instagram, LinkedIn distribution

RMM

Atera

Remote monitoring, device inventory, ticket sync

Email

InMotion/cPanel

Agent email pipes, SMTP relay, push-based ingestion

LLM

Anthropic Claude

Haiku (fast) + Sonnet (complex) via shared client

SLA Commitments

Measured. Accountable. Transparent.

Every metric tracked in real-time and available to clients via the dashboard. No vanity numbers — real operational targets.

Metric
Target
Detail
Platform Uptime
99.9%
GCP Cloud Run + Firebase Hosting SLA-backed
Agent Response (Email)
< 60 seconds
From email receipt to triage + acknowledgment
Agent Response (Phone)
< 5 seconds
Joshua answers inbound calls in real-time
Ticket Triage
< 2 minutes
Classification, routing, and agent assignment
Escalation to Human
< 15 minutes
Critical items flagged via Telegram with full context
Video Pipeline
< 15 minutes
Script → TTS → render → post-production → publish

Ready for a technical deep-dive?

Schedule a 30-minute architecture walkthrough with our engineering team. We'll show you the live system, answer your integration questions, and discuss your security requirements.