Skip to content

JarvisCore Enterprise

JarvisCore is open source under Apache 2.0. You can self-host it, run it on any cloud, and operate it entirely on your own infrastructure — forever, for free.

JarvisCore Enterprise is a managed deployment service. Prescott Data runs, operates, and supports the full JarvisCore stack on your behalf — the same way MongoDB Atlas runs MongoDB or Redis Cloud runs Redis. Your team builds agent systems; we handle the infrastructure that runs them.

OSS vs. Enterprise

JarvisCore OSS JarvisCore Enterprise
Who runs it You Prescott Data
Deployment Self-managed Fully managed
Licensing Apache 2.0 — free Commercial agreement
Uptime SLA — 99.9% guaranteed
Support Community (GitHub Issues) Dedicated engineering + SLA
Security hardening Framework defaults Enterprise-grade (see below)
Data isolation You configure it Enforced at the infrastructure layer
Backup & DR You configure it Automated, cross-region, RPO < 1h
Agent governance None Policy engine, rate limits, allow/deny lists
LLM flexibility Any provider via config Bring-your-own LLM, on-prem model support
Professional services — Onboarding, architecture review, migrations

What Enterprise Covers

Managed Deployment

Prescott Data provisions, configures, and operates the complete JarvisCore stack — agents, mesh, Redis, blob storage, and observability — on your chosen cloud (AWS, Azure, GCP) or in your private network.

  • Zero-ops onboarding: your team connects to a running system, not a setup guide
  • Automated patching, dependency upgrades, and rollbacks
  • Capacity planning and horizontal scaling handled on your behalf
  • Private networking, VPC peering, and on-premises deployment options
  • Kubernetes-native deployment with Helm charts and an operator

Typical deployment topology:

Your Infrastructure (VPC / Private Network)
┌───────────────────────────────────────────────────────────┐
│                                                           │
│   ┌─────────────┐    ┌─────────────┐    ┌─────────────┐  │
│   │  Agent Mesh │    │  Workflow   │    │  Nexus OSS  │  │
│   │  (K8s Pods) │←───│  Engine     │    │  (Auth GW)  │  │
│   └──────┬──────┘    └─────────────┘    └─────────────┘  │
│          │                                                │
│   ┌──────▼──────┐    ┌─────────────┐    ┌─────────────┐  │
│   │  Redis      │    │  Blob       │    │  SIEM /     │  │
│   │  (HA Pair)  │    │  Storage    │    │  Observ.    │  │
│   └─────────────┘    └─────────────┘    └─────────────┘  │
│                                                           │
│   Managed by Prescott Data  ·  Your data never leaves    │
└───────────────────────────────────────────────────────────┘

Uptime SLA

Enterprise deployments carry a 99.9% monthly uptime SLA (≤ 43.8 minutes unplanned downtime per month), with the following terms:

Metric Commitment
Monthly uptime ≥ 99.9%
Measurement window Calendar month, per region
SLA credit Pro-rated service credit if missed
Exclusions Scheduled maintenance (pre-announced ≥ 48h), force majeure, customer-caused outages
Status page Provided — real-time and historical incident data

Full SLA terms are included in the commercial agreement.

Backup & Disaster Recovery

Parameter Value
Recovery Point Objective (RPO) < 1 hour
Recovery Time Objective (RTO) < 4 hours
Backup frequency Continuous (Redis AOF + hourly snapshots)
Backup retention 30 days rolling
Cross-region replication Available on Professional and Enterprise plans
Failover Automated — no manual intervention required

Backups are encrypted with your tenant key and stored in isolated, access-controlled storage separate from the primary deployment.

Security & Authentication

  • SSO via SAML 2.0 or OIDC, federated against your existing identity provider
  • SCIM provisioning and automatic deprovisioning
  • Role-based access control (RBAC) across agents, tools, data sources, and workflows
  • Secrets management via HashiCorp Vault, AWS KMS, Azure Key Vault, or GCP KMS
  • Encryption at rest and in transit; bring-your-own-key (BYOK) option
  • Audit-grade event logs, tamper-evident, exportable to your SIEM

Tenant Isolation

Enterprise enforces strict separation at every layer for multi-tenant deployments serving multiple business units or customers.

  • Workspace-level partitioning: each tenant's agents, workflows, and memory are fully separated
  • Network isolation: no shared data paths between tenants
  • Per-tenant encryption keys
  • Redis namespaces and blob storage prefixes enforced at the platform layer — not in application code
  • Verified isolation documentation available as part of the security review pack

Agent Governance & Policy Controls

Enterprise adds a platform-level policy engine that sits above your agent code. This gives platform and security teams controls that don't require modifying agent implementations.

  • Rate limiting — per-agent, per-workflow, and per-tenant request caps with configurable burst allowances
  • Capability allow/deny lists — restrict which tools, APIs, and data sources each agent role can access
  • LLM provider policy — enforce which LLM providers and model versions agents may use; block unapproved models
  • Output filtering — platform-level PII redaction and content policy enforcement on all agent outputs before they leave the mesh
  • Workflow approval gates — require human sign-off before specific workflow steps execute in production
  • Cost guardrails — per-workspace LLM token budget with configurable alerting and hard caps

LLM Flexibility

Enterprise does not lock you into a single AI provider. You bring the models your team has approved.

  • Use any LLM via the existing JarvisCore provider config (OpenAI, Anthropic, Mistral, Cohere, and others)
  • Bring-your-own-model — connect to on-premises or VPC-hosted models (Ollama, vLLM, Azure OpenAI private endpoint)
  • Per-agent model routing — different agent roles can use different models and providers
  • Model fallback chains — if a primary provider is unavailable, agents fail over to a configured backup automatically
  • All LLM calls are logged, attributable, and subject to your governance policy

Data Privacy & Compliance

  • PII detection and redaction on agent inputs, outputs, and memory writes
  • Field-level encryption for sensitive workflow data
  • Configurable data residency: choose the region where data is stored and processed
  • Retention policies and deletion workflows, including right-to-delete (GDPR Art. 17)
  • Full provenance metadata on all memory reads and writes
  • Compliance documentation available under NDA: SOC 2 Type II report, architecture overview, threat model

Observability & Tracing

  • OpenTelemetry export — metrics, traces, and logs — to your existing stack (Datadog, Grafana, Splunk, and others)
  • SIEM-ready audit streams with tamper-evident event records
  • Long-retention, searchable workflow traces with replay tooling
  • Prometheus metrics aggregated across all nodes and exportable to your monitoring platform

Professional Services

Beyond the managed platform, Prescott Data offers scoped professional services for teams that need hands-on help:

Service Description
Onboarding programme Guided migration from OSS to Enterprise: infra hand-off, agent review, integration setup (2–4 weeks)
Architecture review A senior JarvisCore engineer reviews your agent design, workflow DAGs, and memory strategy and provides a written recommendations report
Custom integrations Prescott Data builds and maintains integrations with internal systems (data warehouses, internal APIs, observability stacks)
Training Hands-on workshops for your engineering team covering agent design patterns, workflow modelling, and production operations

Professional services are scoped and priced separately from the managed platform subscription.

Support Tiers

Plan Response time Coverage Included
Standard Next business day Business hours GitHub + email, documentation
Professional 4 hours 24 × 5 + Named engineer, Slack/Teams channel
Enterprise 1 hour 24 × 7 × 365 + Quarterly architecture reviews, on-call escalation

All paid plans include a named support engineer and a dedicated Slack or Teams channel.

Pricing Model

JarvisCore Enterprise is priced on a subscription basis — a flat annual fee based on deployment size (number of active agents and workflow volume), not per-seat or per-API-call. This means your costs are predictable as you scale agent usage, and you don't get billed every time an agent completes a workflow step.

Pricing is scoped per engagement. Factors include:

  • Number of active agent nodes
  • Monthly workflow execution volume
  • Cloud region and redundancy requirements
  • Support tier
  • Professional services scope

[!NOTE] Contact us with your cloud provider preference, approximate agent count, and any compliance requirements. We'll prepare a scoped proposal within 2 business days.

Who Enterprise Is For

Enterprise is the right choice when any of these apply:

  • You don't want to run infrastructure. Your team should build agent systems, not operate Redis clusters, manage blob storage, or debug Kubernetes rollouts.
  • You have compliance requirements. SOC 2, GDPR, HIPAA, or internal data governance policies require audited, isolated, documented deployments that pass security review.
  • You need a guaranteed uptime SLA. Your agent workflows are in the critical path of production systems and cannot tolerate unplanned downtime.
  • You operate at multi-tenant scale. Multiple business units or customers need strict data separation without the cost of running completely separate deployments.
  • You need a support SLA. When a production incident happens, you need a guaranteed response time and an engineer who knows your deployment — not a GitHub issue queue.
  • You need model and vendor governance. Your organisation requires policy controls over which LLMs agents can call, with auditable records of every call.

How Teams Typically Adopt

  1. Build on JarvisCore OSS. Every framework primitive — agents, memory, orchestration, HITL, Nexus — is available and unrestricted. Validate your architecture on your own infrastructure first.
  2. Move to Enterprise when operational overhead, compliance requirements, or uptime SLAs become the constraint — not the framework's capabilities.
  3. Onboarding typically takes 1–2 weeks from a signed agreement to a fully running managed environment.

Security Review Pack

The security review pack is available to qualified enterprise evaluators under NDA. It includes:

  • Architecture overview and data flow diagrams
  • Threat model and mitigations
  • SOC 2 Type II report (latest period)
  • Penetration test summary (latest annual test)
  • Encryption key management documentation
  • Shared responsibility model

Request the pack via the email below and reference "security review" in your subject line.

Get In Touch

Email: jarviscore-enterprise@prescottdata.io

Include your cloud provider preference, approximate agent count, and any specific compliance framework requirements in your first message — it helps us prepare the right materials and route your enquiry to the right team.

We respond to all enterprise enquiries within one business day.