Overview¶
How it works¶
Every haagsman.ai product follows the same pattern:
- Input — You send data (documents, emails, transcripts, invoices) via REST API
- Process — The product uses AI to analyze, extract, or generate
- Output — You get structured results (answers, categories, action items, drafts)
All products share a common core library that handles authentication, encryption, monitoring, and LLM communication. This means consistent behavior, security, and management across your entire deployment.
Deployment options¶
| Option | Best for | Complexity |
|---|---|---|
| Docker Compose | SMBs, single-server deployments | Simple |
| Kubernetes / Helm | Enterprise, multi-node, auto-scaling | Moderate |
| Cloud templates | AWS, GCP, Azure managed deployments | Moderate |
| Air-gapped (Ollama) | Regulated industries, no internet | Simple |
System requirements¶
Minimum (single product)¶
- 2 CPU cores
- 2 GB RAM
- 10 GB disk
- Docker 24+
Recommended (full suite)¶
- 8 CPU cores
- 16 GB RAM
- 100 GB SSD
- Docker 24+ with Compose v2
LLM requirements¶
- Claude or OpenAI: Internet access to API endpoints
- Ollama (air-gapped): GPU recommended (NVIDIA with 8GB+ VRAM), 16GB+ system RAM
Security at a glance¶
- All data encrypted at rest (AES-256-GCM)
- All traffic encrypted in transit (TLS 1.3)
- JWT + API key authentication with role-based access control
- GDPR-compliant: built-in data export and deletion endpoints
- NIST 800-190 container security guidelines followed
- Non-root containers, read-only filesystems, no shell in production images
- Full audit logging of every API call