Back to Dashboard | Dedicated Domain Manual
Dedicated Domain Tier

Dedicated Domain

User Manual

Complete guide for the Dedicated Domain tier. Your own isolated CPU cluster and managed database at $2.00 per 1M tokens + $2,500/mo base fee. Includes RTX 4000 Ada GPU, continual learning, and custom domain gates.

Getting Started

1. Create Your Account

Visit solacesentry.com/signup to create your account. You will need a valid email address and a password that meets our security requirements.

2. Choose the Dedicated Domain Plan

Select "Dedicated Domain" from the pricing page. Choose one or more of the 25 safety domains. Your plan includes an RTX 4000 Ada GPU -- no separate add-on needed.

3. Complete Payment & Provisioning

After payment, you will be taken through the Setup Wizard (see next section) to configure your dedicated infrastructure. Provisioning typically takes 5-15 minutes.

4. Get Your API Key

After provisioning completes, your API key will be available in your dashboard under API Keys. 

sk_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

Important: Keep your API key secret. Do not share it in public repositories, client-side code, or logs. If compromised, rotate it immediately from your dashboard.

Setup Wizard

After payment, the Setup Wizard guides you through configuring your dedicated infrastructure. The wizard runs automatically and provisions your isolated environment.

1

Domain Configuration

Confirm your selected safety domains and configure domain-specific parameters such as risk tiers (critical, elevated, standard).

2

Infrastructure Provisioning

Your dedicated CPU cluster and managed database are provisioned automatically. A status indicator shows progress in real time. This typically takes 5-15 minutes.

3

SSH Key Upload (Optional)

Upload your SSH public key for direct infrastructure access. You can also add this later from the Entitlements page.

4

API Key Generation

Your production API key is generated and displayed. Copy it immediately as it will only be shown once in full.

Ready

Your dedicated infrastructure is live. You can start submitting observations and running inference immediately.

Your Dedicated Infrastructure

The Dedicated Domain tier provides your own isolated infrastructure. Unlike the Shared Inference tier, your workloads run on dedicated hardware that is not shared with other tenants.

Dedicated CPU Cluster

Isolated compute resources dedicated exclusively to your workloads. No noisy neighbors.

Managed Database

Your own PostgreSQL database with automated backups, connection pooling, and managed migrations.

Continual Learning

Your models improve over time with your data using EWC-based continual learning. No catastrophic forgetting.

Data Isolation

Your data never leaves your dedicated infrastructure. Separate network, compute, and storage from all other tenants.

Infrastructure Status

Your infrastructure status is visible on your dashboard. The status indicator shows:

  • Ready Infrastructure is live and accepting requests
  • Provisioning Infrastructure is being set up (5-15 minutes)
  • Error An issue occurred -- contact support

API Authentication

All API requests must include your API key in the Authorization header using the Bearer token scheme. Your API endpoint is the same for all tiers: 

Authorization: Bearer sk_live_your_key_here

Base URL: https://api.solacesentry.com

Key Prefixes

sk_live_ Production key. All requests are billed and routed to your dedicated infrastructure.
sk_test_ Test key. Requests are free, still routed to your dedicated infrastructure for realistic testing.
sk_dev_ Development key. Local development with mock responses.

Submitting Observations

Observations are the data points you submit for violation detection. On the Dedicated Domain tier, observations are processed on your isolated infrastructure for maximum security and performance.

Using curl

curl -X POST https://api.solacesentry.com/v1/projects/{project_id}/observations \
  -H "Authorization: Bearer sk_live_your_key_here" \
  -H "Content-Type: application/json" \
  -d '{
    "payload": {
      "transaction_amount": "150000.00",
      "account_type": "corporate",
      "risk_flag": "true",
      "domain": "financial"
    }
  }'

Using Python (requests)

import requests

url = "https://api.solacesentry.com/v1/projects/{project_id}/observations"
headers = {
    "Authorization": "Bearer sk_live_your_key_here",
    "Content-Type": "application/json"
}
payload = {
    "payload": {
        "transaction_amount": "150000.00",
        "account_type": "corporate",
        "risk_flag": "true",
        "domain": "financial"
    }
}

response = requests.post(url, json=payload, headers=headers)
print(response.json())

Running Inference

Call the inference endpoint to run violation detection on accumulated evidence. On the Dedicated Domain tier, inference runs on your dedicated hardware with continual learning improvements applied. 

curl -X POST https://api.solacesentry.com/v1/projects/{project_id}/infer \
  -H "Authorization: Bearer sk_live_your_key_here" \
  -H "Content-Type: application/json"

Response Format

{
  "classification": "concern",
  "narrative": "Large corporate transaction ($150,000) with risk flag set to true. Transaction amount exceeds standard thresholds for the financial domain. Recommend manual review before processing.",
  "decision_trace": {
    "sparse_gate": "passed",
    "judge_verdicts": [ ... ],
    "tribunal_outcome": "concern"
  },
  "evidence_state": {
    "current_weight": 0.62,
    "observation_count": 5
  }
}

Understanding Results

Classification Levels

VETO Critical violation -- stop and escalate

Immediate human review required. Proceeding would pose unacceptable risk.

CONCERN Potential issue -- review recommended

Patterns warrant attention. Review narrative and evidence to determine action.

APPROVE No violations -- continue operations

Observations fall within expected parameters. No safety issues detected.

Every response includes a narrative grounded in evidence (INV-8), a decision_trace for full explainability, and the current evidence_state. SolaceSentry never operates as a black box.

Evidence & Expectations

Evidence State

Evidence accumulates and never decays (INV-2). Retrieve your project's current evidence state at any time: 

curl -X GET https://api.solacesentry.com/v1/projects/{project_id}/evidence \
  -H "Authorization: Bearer sk_live_your_key_here"

Setting Expectations

Define expected data bounds. Violations of expectations strengthen evidence weight: 

curl -X POST https://api.solacesentry.com/v1/projects/{project_id}/expectations \
  -H "Authorization: Bearer sk_live_your_key_here" \
  -H "Content-Type: application/json" \
  -d '{
    "expectations": [
      { "field": "transaction_amount", "min": "0", "max": "50000", "unit": "usd" }
    ]
  }'

Python SDK

pip install solace-sentry
import asyncio
from solace_sentry.sdk import SolaceSentryClient

async def main():
    client = SolaceSentryClient(
        api_key="sk_live_your_key_here",
        base_url="https://api.solacesentry.com"
    )

    # Submit an observation
    obs = await client.observations.create(
        project_id="proj_abc123",
        payload={
            "transaction_amount": "150000.00",
            "domain": "financial"
        }
    )

    # Run inference
    result = await client.inference.create(project_id="proj_abc123")
    print(result.classification)  # "veto", "concern", or "approve"
    print(result.narrative)

asyncio.run(main())

Using the Interpreter

The Interpreter (Clinical Reasoning Workbench) is a natural language interface in your dashboard under each entitlement. It supports all 25 safety domains with 12 query intents including assess_risk, explain_decision, compare_scenarios, list_violations, show_evidence, trace_decision, suggest_action, summarize_state, query_history, check_compliance, forecast_trend, and validate_data.

How to Access

  1. Log in at solacesentry.com/client/dashboard
  2. Navigate to Entitlements
  3. Click Interpreter on any active entitlement
  4. Type your question in natural language

Example queries: "What violations have been detected?", "Why was the last inference vetoed?", "Show me the current evidence state", "Compare the last two scenarios"

Continual Learning

Dedicated Domain Exclusive

Continual learning is exclusive to the Dedicated Domain and Enterprise Security tiers. Your models improve over time using your own data without catastrophic forgetting.

How It Works

SolaceSentry uses Elastic Weight Consolidation (EWC) to continuously learn from your domain data while preserving previously learned knowledge:

  • Fisher Information Matrix -- identifies which model parameters are most important for existing knowledge
  • EWC Regularization -- penalizes changes to important parameters during new learning
  • Replay Buffer -- stores representative past examples to prevent drift
  • Progressive Network -- adds capacity for new domain knowledge without overwriting
  • Repetition Validator -- ensures new learning does not cause repetitive or degraded outputs

What Triggers Learning

The ContinualLearningWorker runs as a background process on your dedicated infrastructure. It automatically processes new observations and inference results to improve domain-specific accuracy. Learning happens asynchronously and does not affect inference latency.

Safety Guarantees

All 8 hard invariants remain enforced at all times during continual learning. The system cannot learn to bypass safety gates, decay evidence, or violate any invariant. If a learning update would degrade safety metrics below acceptance thresholds, it is automatically rolled back.

Custom Domain Gates

Domain gates are configurable rules that apply domain-specific logic during inference. On the Dedicated Domain tier, you can customize these gates to match your organization's specific policies and thresholds.

Available Domain Gates

SolaceSentry ships with 5 built-in domain gate types. Each can be configured per-domain:

Healthcare Gate Clinical thresholds, drug interaction checks, vital sign bounds
Financial Gate Transaction limits, fraud patterns, regulatory compliance
Autonomous Gate Safety margins, sensor validation, emergency thresholds
Therapy Gate Treatment protocol adherence, dosage limits
Legal Gate Regulatory requirements, compliance deadlines

To configure custom domain gates, contact your support channel or use the dashboard settings under your entitlement configuration.

Dedicated GPU (Included)

Your Dedicated plan includes an NVIDIA RTX 4000 Ada GPU for accelerated inference and continual learning. No add-on required -- GPU compute is part of your base subscription.

RTX 4000 Ada (20GB VRAM)

Dedicated GPU for your isolated inference workload.

  • 20GB GDDR6 memory
  • 5-10ms inference latency
  • FlashAttention-2 enabled
  • Continual learning on your domain data
  • 10M tokens/mo included

GPU infrastructure is provisioned automatically when your Dedicated plan is activated. Provisioning takes approximately 5-10 minutes. For H100 GPU (Enterprise tier), upgrade to Enterprise Security.

SSH Access

The Dedicated Domain tier provides SSH access to your infrastructure for advanced configuration, log inspection, and direct debugging.

Setting Up SSH

  1. Navigate to Entitlements in your dashboard
  2. Click Manage on your Dedicated Domain entitlement
  3. Upload your SSH public key (ed25519 or RSA 4096-bit recommended)
  4. Your SSH endpoint and port will be displayed after key upload

Connecting

ssh -i ~/.ssh/your_key solace@your-instance.solacesentry.com -p 2222

Note: SSH access is for inspection and configuration only. Core system files and invariant-enforcing code are read-only and cannot be modified via SSH.

Seat Management

The Dedicated Domain tier includes up to 10 seats. Each seat is a user who can access the dashboard, manage API keys, and use the Interpreter.

Managing Team Members

  1. Navigate to Profile in your dashboard
  2. Click Team Members
  3. Invite team members by email
  4. Assign roles: Admin, Developer, or Viewer
Role Dashboard API Keys Interpreter Billing
Admin Yes Yes Yes Yes
Developer Yes Yes Yes No
Viewer Yes No Yes No

Safety Domains

All 25 safety domains are available across 6 categories:

Healthcare

healthcare_ops

clinical

pharma

lab

Financial

revenue

financial

insurance

claims

fraud

Legal & Regulatory

legal

regulatory

government

Cyber & Security

cybersec

threat

incident

ai_governance

Industrial

manufacturing

supply_chain

energy

infrastructure

Transport & People

aviation

autonomous

safety_eng

hr

Hard Invariants

8 invariants enforced at all times -- even during continual learning:

1. Sparse Gate

Fast-path bypass for trivial observations

2. No-Decay Evidence

Evidence weights never decrease

3. Lazy Staleness

Stale evidence detected lazily at read time

4. Fast Gate Before Planning

Planning only invoked if necessary

5. Planning Gated

Crisis check before any planning

6. Max 2 Narrative Attempts

Fallback used if generation fails twice

7. Record Immutability

Records cannot be modified after creation

8. Narrative Reads Record Only

Narratives always grounded in recorded evidence

Rate Limits

Priority rate limits on dedicated infrastructure:

Endpoint Rate Limit Burst
Observations 300 requests/min 50
Inference 300 requests/min 50
Evidence / Expectations 600 requests/min 100
Health Check 1200 requests/min 200

Rate limits on the Dedicated Domain tier are 5x higher than Shared Inference. A 429 response includes a Retry-After header.

Billing & Usage

Pricing

$2.00 / 1M tokens

+ $2,500/mo base fee for dedicated infrastructure

What Is Included

  • Dedicated CPU cluster
  • Managed PostgreSQL database
  • Continual learning (EWC)
  • 10 team seats
  • SSH access
  • Priority rate limits (300 rpm)
  • Slack + email support

GPU Included

RTX 4000 Ada GPU is included with your Dedicated plan at no extra cost. For H100 performance, upgrade to Enterprise Security ($5,500/mo).

API Reference

Base URL: https://api.solacesentry.com

Method Endpoint Description
POST /v1/projects/{project_id}/observations Submit an observation
POST /v1/projects/{project_id}/infer Run violation inference
GET /v1/projects/{project_id}/evidence Get current evidence state
GET /v1/projects/{project_id}/expectations Get expectations
POST /v1/projects/{project_id}/expectations Set expectations
GET /v1/health Health check

Support

Email Support

support@solacesentry.com

Priority response: within 4 business hours

Slack Support

Dedicated Slack channel for your team

Set up during provisioning

FAQ

Can I upgrade to Enterprise Security later?

Yes. Your data and infrastructure will be migrated. Contact support to coordinate the transition.

How long does provisioning take?

Typically 5-15 minutes for full infrastructure including GPU (RTX 4000 Ada).

Is my data used to train models for other tenants?

No. Your data stays on your dedicated infrastructure. Continual learning only improves your own models. No data is shared across tenants.

Can I add more than 10 seats?

The Dedicated plan includes up to 50 seats. For unlimited seats, consider the Enterprise Security tier or contact sales for custom arrangements.

Is the Dedicated Domain tier suitable for HIPAA data?

While data is isolated, the Dedicated Domain tier does not include HIPAA compliance, BAA, or SOC 2 certification. For regulated data, use the Enterprise Security tier.

What happens if I cancel?

You will retain access until the end of your billing period. Data export is available via the dashboard. After the period ends, your infrastructure is decommissioned.