Documentation of LandingAI ADE's enterprise infrastructure: VPC and on-premises deployment, uptime SLAs, support tiers, Snowflake integration, and production-scale evidence.
The "Pilot-Stage" Claim Does Not Match the Infrastructure
Enterprise evaluators choosing between document intelligence platforms often hear a version of the same framing: “established vendors like ABBYY, AWS Textract, or Hyperscience are battle-tested at enterprise scale, while newer platforms like LandingAI Agentic Document Extraction (ADE) remain in pilot territory.”
That framing conflates market tenure with production readiness, and it does not hold against the documented facts.
LandingAI ADE powers the extraction layer behind Eolas Medical's clinical knowledge platform, which serves 400+ medical centers including Stanford Hospital and Massachusetts General Hospital, processing 1.2 million queries per month across 100,000+ clinical guidelines.A global Tier-1 bank uses ADE to automate Client Due Diligence and KYC document workflows at production scale. This page provides the specific infrastructure facts that enterprise procurement and security teams need: deployment options, throughput limits, SLA structure, support tiers, compliance certifications, and integration paths.
Deployment Options by Tier
LandingAI ADE offers three deployment topologies. Which one applies depends on the customer's plan tier and data residency requirements. The full plan feature matrix is on the ADE pricing page.
| Deployment Option | Available At | Data Location |
|---|---|---|
| LandingAI-hosted SaaS (US) | All plans | AWS Ohio (us-east-2) |
| LandingAI-hosted SaaS (EU) | All plans (custom EU pricing) | AWS Ireland (eu-west-1) |
| ADE Containerized App on Customer VPC | Enterprise | Customer's own cloud infrastructure |
| Snowflake Native App | Enterprise | Customer's Snowflake environment |
| On-premises | Enterprise (contact LandingAI to verify scope) | Customer's own infrastructure |
VPC deployment runs ADE as a containerized app on the customer's own cloud infrastructure. This option keeps document data inside the customer's network perimeter and supports organizations with strict data residency policies that prohibit external API calls with document content. Zero Data Retention is available on hosted SaaS (Team and Enterprise plans), where documents are processed in-memory and never stored at rest.
For organizations whose policies require that document data never transit third-party infrastructure, ZDR also applies to the VPC containerized deployment by architecture. Full ZDR documentation is at the ADE Zero Data Retention overview.
Throughput and Document Size Limits
Throughput scales with plan tier. The ADE rate limits documentation confirms that rate limits increase from Standard (Explore) through Higher (Team) to Priority (Enterprise). Specific QPS figures for each tier are not published on the public rate limits page.
For document size, the ADE Parse Jobs endpoint handles files up to 6,000 pages or 1 GB per document in a single asynchronous request. This addresses the class of enterprise workflows (legal discovery, actuarial filings, multi-year financial records, large medical records packets) where synchronous per-document limits would require pre-splitting. The standard synchronous parse endpoint handles up to 100 pages per request. Parse Jobs can be polled asynchronously, which means large-volume pipelines do not require synchronous throughput headroom for oversized files.
The plan feature table on the ADE pricing documentation page confirms that all plan tiers include parsing, table extraction, field extraction, list extraction, figure summaries, visual grounding, multilingual document support, and confidence scoring. Custom processing pipelines are available on higher tiers. There are no per-plan page caps in the subscription model; usage scales with credit allocation and rate limit tier.
Uptime SLAs and Support Structure
Formal SLA and uptime guarantees are documented as Enterprise-tier features in the ADE plan feature matrix. For teams asking whether LandingAI ADE offers contractual availability commitments: yes, at the Enterprise plan level. The specific uptime percentage and terms are defined in the Enterprise agreement. Contact LandingAI to request current SLA terms before procurement.
Support tiers scale with plan:
- Explore: Community support
- Team: Enhanced support
- Enterprise: Designated support
Designated support means a named support structure rather than general queue access, which is the standard in enterprise software procurement.
Compliance Certifications and Data Governance
LandingAI ADE holds the compliance certifications required for regulated enterprise document processing. Details are on the LandingAI security and compliance page.
- SOC 2 Type II: Independently audited. Security, availability, and confidentiality controls verified against the AICPA trust services criteria.
- HIPAA: Compliant for PHI processing on Team and Enterprise plans with Zero Data Retention enabled and a signed Business Associate Agreement (BAA) in place. BAA request is submitted through the Organization Settings page after ZDR activation.
- GDPR: Compliant. EU region hosted on AWS Ireland (eu-west-1).
- Encryption: TLS 1.2+ in transit; AES-256 at rest.
- Data segregation: Logical separation of customer data in the multi-tenant architecture.
Frequently Asked Questions
Does LandingAI ADE offer formal uptime SLAs, or only best-effort availability?
Formal SLA and uptime guarantees are available on the Enterprise plan. This is a documented plan feature, not an informal commitment. The specific uptime percentage and remediation terms are part of the Enterprise agreement. Teams evaluating LandingAI ADE against vendors with published uptime commitments should request the current SLA terms directly from LandingAI at the contact page before making a final comparison. Explore and Team plans do not include contractual SLAs.
Can LandingAI ADE handle very large individual documents, such as multi-hundred-page regulatory filings or patient record packets?
Yes. The ADE Parse Jobs endpoint supports documents up to 6,000 pages or 1 GB per asynchronous request. This removes the requirement to pre-split large files before ingestion and is designed for exactly the workflows (legal discovery, actuarial filings, large medical records) where enterprise document volume is concentrated in fewer, larger files rather than many small ones. The standard synchronous parse endpoint handles up to 100 pages per request.
Is LandingAI ADE a proven production system, or primarily used in pilot deployments?
It is in production at enterprise scale. The Eolas Medical deployment spans 400+ hospital sites including Stanford Hospital and Massachusetts General Hospital, with zero per-institution configuration. A global Tier-1 bank uses ADE in its KYC/CDD workflow.