Enterprise Deployment, Your Way
Bring Your Own Cloud (BYOC) or use Across-managed infrastructure - with Bring Your Own Model (BYOM) support and full data isolation for long-horizon agents with memory.
BYOC-first
Run entirely inside your AWS, Azure, or GCP VPC. You control IAM, networking, and logs.
BYOM-ready
Use your approved LLMs or Across-managed models: enterprise-grade and compliant - no training on customer data.
Isolated by design
Dedicated Kubernetes deployments per customer. Zero-trust architecture with complete data isolation.
Deployment Options
Bring Your Own Cloud (BYOC) or Managed Service
Bring Your Own Cloud (BYOC)
Best for: Security-sensitive enterprises
- • Deployed into your AWS / Azure / GCP VPC
- • You control networking, IAM, logging, and storage
- • No data egress outside your environment
- • Works with your existing security tooling
Across-Managed Service
Best for: Fast pilots and early deployments
- • Dedicated, isolated environment managed by Across
- • Encrypted at rest and in transit
- • Enterprise access controls and audit logging
- • Easy path to migrate to BYOC later
Your Data Stays Isolated
- •Tenant isolation at compute, storage, and model layers
- •Permission-preserving memory (RBAC / PBAC inherited from source systems)
- •Configurable data retention and deletion policies
- •Full audit trail of access, inference, and actions
- •No training on customer data
Bring Your Own Model
- •Azure OpenAI, internal LLMs, or other enterprise-approved models
- •If using Across hosted modles: Isolated enterprise agreements, no training on your data
- •Across agents use different models for fast decisions vs deep reasoning
User Experience & Surfaces
- •Primary: Microsoft Teams, Slack, or Webex (verified apps)
- •Next-best actions show up within tools that your team already uses, e.g. Outlook, Word, Excel
- •Pro-active and fits into your existing workflows
POC → Pilot → Production
From Deployment to Measurable Impact
POC (4–6 weeks)
Validate fit, configure 2–3 micro-workflows
Pilot
Measure uplift in agreed metrics (cycle time, conversion, response rate)
Production
Expand workflows, teams, and models