Enterprise AI Orchestration — Governed · Auditable · Azure-Native
AI-as-a-Service (AIXaaS™) is Inflexis's enterprise AI orchestration platform. It deploys in your tenant environment or can be hosted on ours as a iSaaS product, connects to your documents and operational data, and gives every role in your organization governed, auditable AI access — with a full change-control trail, role-based permissions, and zero IP exposure to third parties.
Unlike a chatbot or a simple API wrapper, AIXaaS™ is a complete orchestration system:
Your Documents + Data
↓
Universal Ingestor (20+ file types)
↓
PII Masking → Compliance Scan → Chunk → 4-Tier Store
↓
8-Role RBAC + Namespace Governance
↓
Cost-Aware Model Routing (6 LLM providers, cheapest first)
↓
Grounded Response + Source Citations + ADR Audit Record
↓
Your Team (role-appropriate answer, in their namespace)
Every step is logged. Every agent instruction change is version-controlled. Every provider has a fallback.
| Platform Overview | What AIXaaS™ is, the problems it solves, and core design principles |
| Architecture Reference | Full system topology, storage tiers, security model |
| Deployment Guide | Azure Container Apps CI/CD, rollback, custom domains, health checks |
| Azure Services Reference | All Azure resources, configuration, CLI commands |
| LLM Provider Configuration | 6-tier routing, role policies, data residency, fallback behavior |
| API Reference | All endpoints with curl examples, request/response schemas |
| MCP Server | Claude Desktop integration — 8 tools exposed |
| RAG Pipeline | Ingest → chunk → store → retrieve — all 5 pipeline stages |
| Compliance Engine | 23 frameworks, zero-token detection, framework reference |
| Meeting Intelligence | Transcription, SPICED scoring, delivery pipeline |
| ADR Change Control | Agent governance, diff review, archive/rollback |
| Ingesting Documents | Upload documents, batch ingest, supported file types |
| Roles & Namespaces | Configure RBAC, namespace isolation, sector governance |
| Multi-Tenant Setup | Per-client namespace isolation, white-label deployment |
📊 System Topology
┌────────────────────────────────────────────────────────────────────┐
│ AIXaaS™ / MAO Platform │
│ Azure Container Apps │
├──────────────────┬──────────────────┬──────────────────────────────┤
│ INGEST PIPELINE │ KNOWLEDGE BASE │ AGENT LAYER │
│ │ │ │
│ PDF DOCX PPTX │ T1: Keyword │ 8-Role RBAC │
│ XLSX MP3 MP4 │ T2: Qdrant/vec │ Compliance Engine (23 fw) │
│ YouTube Web │ T3: Pinecone │ Meeting Intelligence │
│ Markdown RTF │ T4: Azure Search│ ADR Change Control │
│ CSV JSON HTML │ │ Semantic Cache (30-60% ↓) │
│ │ 4-Tier Hybrid │ Token Budget Guardrails │
│ → Parser Chain │ BM25 + Vector │ 6-Tier Model Router │
│ → PII Mask │ Always Warm │ MCP Server (8 tools) │
│ → Comply Scan │ Disk Persisted │ │
│ → Chunk + Store │ │ │
└──────────────────┴──────────────────┴──────────────────────────────┘
↕ Azure Key Vault · Entra SSO · App Insights ↕
🏷️ Compliance Frameworks (23 — Zero Token Cost)
GDPR HIPAA PCI-DSS SOC 2 NIST CSF ISO 27001 CCPA FedRAMP FISMA FERPA GLBA NERC CIP IEC 62443 NIST SP 800-82 CMMC ITAR EAR SEC 17a-4 FINRA MiFID II Basel III Solvency II DORA
All 23 detected via deterministic pattern matching — no LLM call, no API cost, instant classification.
👤 8-Role RBAC Role based controls with subagents ABAC governance
*Examples*| Role | Min LLM Tier | Namespace Access | Use Case |
|---|---|---|---|
admin |
Tier 3 | All | Full platform control, ADR approvals |
founder |
Tier 3 | All | Executive access, strategic namespaces |
cro |
Tier 3 | Sales + meetings | Pipeline, SPICED scoring, deal data |
marketing |
Tier 2 | Brand + marketing | Campaigns, content, brand sector |
developer |
Tier 2 | Technical + docs | API, codebase, documentation |
member |
Tier 1 | General | Standard knowledge base access |
client |
Tier 2 | Own namespace only | Isolated client data |
Prospects |
Tier 1 | Sandbox only | Demo environment, no production data |
import requests
BASE_URL = "https://app.inflexis.ai"
AUTH = ("your-username", "your-password")
# Search the knowledge base
response = requests.post(f"{BASE_URL}/api/chat", auth=AUTH, json={
"query": "What compliance frameworks apply to our industrial documents?",
"namespace": "compliance",
"top_k": 8,
})
result = response.json()
print(result["response"])
# → Grounded answer with sources, model used, and budget statusSee mao-examples for full scripts and Jupyter notebooks.
AIXaaS™ builds on, integrates with, and complements the best open-source AI infrastructure:
| Project | Stars | Relationship |
|---|---|---|
| anthropics/anthropic-sdk-python |  |
Primary LLM SDK — MAO uses Claude Haiku + Sonnet |
| anthropics/anthropic-cookbook |  |
Reference patterns for agent design |
| modelcontextprotocol/servers |  |
MCP ecosystem — MAO ships its own 8-tool MCP server |
| microsoft/autogen |  |
Multi-agent framework — MAO adds enterprise governance AutoGen lacks |
| langchain-ai/langgraph |  |
DAG orchestration — MAO's pipeline uses same pattern, without the overhead |
| joaomdmoura/crewAI |  |
Role-based agent crews — MAO's RBAC is the enterprise governance layer |
| BerriAI/litellm |  |
LLM routing — MAO's model router adds per-role cost governance |
| qdrant/qdrant |  |
Vector database — MAO's Tier 2 semantic store |
| run-llama/llama_index |  |
RAG framework reference — MAO's hybrid retrieval draws on these patterns |
This is a documentation repository for the AIXaaS™ platform. We welcome:
- Bug reports — broken links, outdated information, incorrect CLI commands → Open an issue
- Documentation improvements — clarity, typos, missing content → Open a PR
- Feature documentation requests — missing docs for a platform capability → Open an issue
See CONTRIBUTING.md for guidelines.
AIXaaS™ is currently in enterprise pilot. Contact us to request access or discuss deployment:
📧 bryan.shaw@inflexis.ai 🌐 inflexis.ai 💼 LinkedIn — Bryan Shaw