Deployment Options

Extend offers three deployment models with different tradeoffs on speed, control, and data residency. All models include enterprise security controls — see Compliance for SOC 2, HIPAA, and GDPR coverage.

Compare deployment models

	Cloud (Extend SaaS)	BYOC	Hybrid
Best for	Moving fast with minimal ops overhead	Data sovereignty, regulated industries, existing cloud commitments	In-country data storage with managed AI infrastructure
Where data lives	Extend-managed US or EU regions	Your cloud account	Documents in your cloud; inference in Extend’s cloud
Infrastructure	Fully managed by Extend	Deployed in your account; Extend manages setup	Split: app/data in your cloud, AI in Extend’s cloud
Time to production	Fastest — accounts provisioned and ready	~2–3 weeks with guided provisioning	Faster setup than full BYOC; lower infra cost
GPU / AI ops	Managed by Extend	You provision GPU quota and cloud AI services	No GPU quota or model management on your end

Cloud (Extend SaaS)

Extend manages all infrastructure, updates, and monitoring. Choose a region when you sign up:

Deployment	Region	API
us1	US	`https://api.extend.ai`
us2	US	`https://api.us2.extend.app`
eu1	EU	`https://api.eu1.extend.ai`

See Deployments for SDK configuration and authentication.

Why teams choose Cloud:

No infrastructure setup — get started immediately
Extend handles scaling, updates, and security patches
Subscription pricing with no infrastructure costs
Proactive monitoring of application health and performance
Inherit Extend’s SOC 2, HIPAA, and GDPR certifications at the platform level

BYOC (Bring Your Own Cloud)

Extend deploys into your cloud account so customer data never leaves your environment. You apply your existing IAM, VPC, and encryption policies while Extend handles deployment automation and application configuration.

Why teams choose BYOC:

Data sovereignty and strict data-residency requirements
Dedicated environment with no shared infrastructure
Leverage committed cloud spend or reserved instances
Flexible update windows scheduled around your operations

What you provision

Extend’s deployment automation handles networking, databases, storage, TLS, and application services. Your team typically provides:

A dedicated cloud project with administrative access and a domain you control
GPU compute quota — L4 instances for layout and reading order; H100 for table extraction and form parsing
Cloud AI services — cloud-native OCR and LLM inference (for example, Vertex AI), with inference staying in your account
DNS delegation — point your domain’s nameservers to the cloud-managed zones Extend creates during deployment
SSO — identity provider integration (Google Workspace, Okta, or Microsoft Entra ID)

Infrastructure components

Deployed in your cloud account:

Extend application — container orchestration
Database and cache — PostgreSQL and Redis
Object storage — documents and files
DNS and TLS — domain and certificates
Identity — SSO via Google Workspace, Okta, or Microsoft Entra ID
GPU compute, OCR, and LLM inference — document AI models in your account
Extend BYOC controller — deploys via service account

Hybrid

Documents and application data stay in your cloud. AI models (GPU compute, OCR, LLM inference) run in Extend’s cloud. Documents are sent to Extend for inference; results return to your environment.

Why teams choose Hybrid:

Meets data-residency rules that allow processing outside the country while keeping storage in-country
No GPU quota, model updates, or AI infrastructure to operate
Faster setup and lower infrastructure cost than full BYOC
Apply your IAM, VPC, and encryption policies to the customer-side deployment

Hybrid deploys the same customer-side components as BYOC (application, database, object storage, DNS, SSO) without provisioning GPU or cloud AI services in your account.