Built for SAP. Runs On Your Infrastructure.
Verporta is a Python/FastAPI backend with a React frontend. It connects to SAP via standard RFC/BAPI interfaces — no SAP add-on, no BASIS team involvement for installation.
4-Layer Intelligence Stack
Each layer is independently scalable. The AI engine runs entirely on your infrastructure — no cloud dependency.
Infrastructure Trust Guarantees
- ✓On-premise / Private cloud
- ✓No SAP data leaves your infrastructure
- ✓Python + FastAPI backend
- ✓Docker / Kubernetes ready
- ✓SAP ECC + S/4HANA compatible
- ✓Rules + ML — not LLM hype
Open-Source Core. Enterprise Ready.
| Layer | Technologies |
|---|---|
| Backend | Python 3.11 · FastAPI · Uvicorn · Celery (async tasks) |
| ML / AI | XGBoost · scikit-learn · ChromaDB · sentence-transformers |
| LLM | Claude API (Anthropic) — root cause explanation & RAG reasoning only |
| SAP layer | pyrfc (RFC connector) · ABAP RFC function modules · Standard BAPIs |
| Frontend | React + Next.js · TypeScript · Tailwind CSS |
| Database | PostgreSQL (metadata + audit) · ChromaDB (vectors) · Redis (job queue) |
| Deployment | Docker · Kubernetes-ready · On-premise or private cloud (AWS/Azure VPC) |
| Monitoring | Built-in health dashboard · Prometheus metrics · Grafana-compatible |
Step-by-Step: From SAP Event to Resolution
SAP IDoc status changes (e.g. status 51 → error). BAdI or RFC polling detects the event.
IDoc control record + status record extracted via RFC. Segment data parsed.
ML classifier assigns: domain, error_type, root_cause_category, confidence_score.
RAG engine queries vector store: finds top-5 similar historical cases.
If confidence > 85% AND pattern in auto-fix registry: execute fix. Log to audit trail.
If confidence < 85%: LLM generates root cause explanation + fix recommendation. Sent to dashboard.
Operator acts or approves auto-fix. All actions logged with timestamp + user + outcome.
Enterprise Security by Design
Data residency
100% on-premise. No SAP data transmitted to external servers.
LLM usage
Only anonymised error descriptions sent to Claude API. PII stripped. Transactional data never leaves.
Authentication
LDAP/AD integration. Role-based access: admin, analyst, read-only.
Audit trail
All AI decisions, auto-fix actions, operator overrides logged with full context.
SAP authorizations
Uses standard SAP RFC user. Minimum required auth objects documented.
Encryption
TLS 1.3 for all internal communication. PostgreSQL encryption at rest.
Three Ways to Deploy Verporta
All options keep your SAP data on your infrastructure. Choose the model that fits your team.
On-Premise Docker
8 vCPU · 32 GB RAM · 500 GB SSD
2–4 hours
1 Linux admin + 1 SAP BASIS (RFC user setup only)
Strict data residency. Air-gapped SAP environments.
Private Cloud (AWS/Azure)
Deployed in customer's own cloud account. Verporta never has access.
1 day
1 DevOps + 1 SAP BASIS
Customers with existing AWS/Azure footprint.
Managed (SaaS-lite)
Single-tenant environment. SAP connection via encrypted tunnel.
2–3 days (tunnel setup)
SAP BASIS for tunnel setup only
Smaller teams without DevOps. AMS providers managing multiple clients.
See Verporta Inside Your SAP Environment
30-minute live demo. We connect to a sample SAP system and show IDoc AI working in real-time. No slides. No pitch deck. Just the product.