Three products. One accountability graph.
From open-source SDK to enterprise governance platform. Each phase delivers a complete, deployable product — built on the same canonical data model.
Capture · Comply · Enforce · Govern
The Providex Platform
Three products. One accountability graph.
RootSign SDK
✓ ShippedTamper-evident provenance logging for your LangGraph or CrewAI pipeline. Apache 2.0.
Install →Providex Comply
Coming soonAgent inventory, framework mapping, and one-click audit reports for compliance teams.
Early Access →Providex Enforce + Govern
Coming soonRuntime policy enforcement, anomaly detection, and cross-platform governance fabric.
Notify Me →The four-phase architecture
Each phase ships a complete product for a specific buyer — and extends the same accountability graph the previous phase built.
Capture
Drop-in instrumentation for production agent pipelines. The open-source SDK that turns any LangGraph, CrewAI, or AutoGen workflow into a tamper-evident Decision and Action stream.
- @rootsign.trace decorator: LangGraph, CrewAI, AutoGen
- Action records: SHA-256 hash-chained, tamper-evident
- Approval records: every HiTL checkpoint outcome logged
- CLI verify: VALID / TAMPERED in under 30 seconds
Comply
The compliance intelligence layer. Agent inventory, regulatory framework mapping, and one-click audit reports — without filing an engineering ticket.
- Agent inventory dashboard: every deployed agent registered
- Session + Decision + Action timeline: the full why + what
- Framework mapping: SOC 2, HIPAA, EU AI Act Article 12/13/14
- One-click audit report: PDF/Word, hash-chain proof included
- Langfuse / Galileo connector: traces become audit evidence
Enforce
Runtime policy enforcement and incident response. Cedar/OPA policies evaluated against every Action, statistical anomaly detection, and bi-directional integrations with the AI security platforms you already run.
- Policy-as-code engine: Cedar/OPA evaluates every Action
- Anomaly detection: statistical baseline, auto-creates Incidents
- Zenity / Rubrik integration: agent discovery enriches Agent entity
- SIEM/SOAR streaming: Incidents flow to Datadog, Splunk, PagerDuty
Govern
The cross-platform governance fabric. Multi-system correlation, policy simulation before live deployment, and GRC connectors that make Providex the platform of record for enterprise AI accountability.
- Cross-system correlation: multi-agent, multi-platform workflows
- Policy simulation: what-if analysis before live deployment
- GRC connectors: Credo AI, Holistic AI, ServiceNow
- Board analytics: override trends, incident rates, posture scores
Open-source SDK. One line. Full provenance.
The RootSign SDK wraps your existing agent tool calls and emits a Decision + Action record for each invocation. Hash-chained, tamper-evident, and indistinguishable from your current pipeline from the outside.
Drop-in decorator
@rootsign.trace wraps any Python function. Works with LangGraph, CrewAI, AutoGen, or custom agents.
< 5ms overhead
Async, batched writes. Your agents won't notice the difference.
Tamper-evident by design
Every Action is SHA-256 hash-chained to its predecessor. A single modification breaks the chain.
HiTL Approval records
Built-in authorization checkpoints. Every approval logged with full Decision context.
import rootsign
from rootsign import session
# Wrap your existing agent tools — one line, no logic changes
async with session(agent_id=agent.agent_id, client=client) as ctx:
tools = rootsign.wrap_tools([query_database, send_email], ctx=ctx)
# your graph runs here — every Decision + Action captured
# Verify the tamper-evident chain
# rootsign verify <session_id>
# VALID ✓ — 47 records, chain intactSeven entities. One accountability graph.
The Providex data model captures not just what agents did, but why they decided to do it — and under whose authority.
Agent
Who deployed the AI, what it's permitted to do.
Session
One workflow run — objective, time window, outcome.
Decision
Why the agent chose this action — the missing layer.
Action
What the agent did — tamper-evident, hash-chained.
Policy
The rule that governs this action, with regulatory refs.
Approval
Who authorized it, with what context, when.
Incident
The investigation record when something goes wrong.
Decision is the entity no other platform captures. It's the difference between a log and an alibi.
Packaging & Pricing
Three editions designed to meet you where you are — from early experimentation to full enterprise governance.
Developer / Team Edition
For platform teams and advanced developers piloting agentic workflows.
- Core SDKs
- Agent inventory
- Session and decision timelines
- Basic policy-as-code checks
- Lightweight approval flows
Faster debugging and safer experimentation before enterprise-wide rollout.
Install SDKEnterprise Standard
For AI governance and risk teams in regulated enterprises.
- All Developer Edition features
- EU AI Act-aligned logging profiles and report templates
- Integrations with GRC, AI governance tools, and major SIEMs
- Role-based access controls
- Data residency options
- Advanced retention policies
Enterprise-grade governance with compliance frameworks built in.
Book a DemoEnterprise Custom
For CISOs and security operations teams with advanced requirements.
- All Enterprise Standard features
- Deeper integrations with AI security platforms
- Enhanced incident workflows
- Correlation with security events
- Custom compliance framework packs (EU AI Act, banking, healthcare)
- Advanced analytics modules
Full-spectrum security and compliance tailored to your organization.
Contact UsHybrid Pricing Model
- •Base platform fee per tenant reflecting compliance value and feature tier
- •Usage-based metering tied to volume of decisions or actions recorded (not prompts alone), aligned with observability pricing practices
- •Add-on bundles for high-value connectors (e.g., EU AI Act pack, banking/healthcare regulatory packs) and advanced analytics modules