Your agents already run across Microsoft, Google, Salesforce, ServiceNow, AWS, and your own builds. Every platform governs its own and quietly prefers it. We bring them onto one record, starting with the major hyperscalers and your own builds, with more on the roadmap.
Meridian is the layer above.
Continuously inventory every agent across vendors, clouds, and BYOA, sanctioned and shadow.
Issue passports, assign owners, and bind agents to your policy envelope before they run.
A single roster, starting at launch with Microsoft agents and your custom stacks.
Apply runtime guardrails, scoped permissions, approvals, and reversible actions.
Stream signed evidence to audit, risk, and compliance systems automatically.
Attribute cost, performance, and outcomes per agent, model, owner, and project.
Everything lands on one record.
Sanctioned, shadow, and custom agents land in a single inventory the moment Discovery finds them. No more hunting across consoles.
Vendor-neutral identity for every agent.
What each agent can touch, enforced at runtime.
Replay any run. Export to your evidence system.
Spend by agent, model, owner, and business unit.
Five building blocks. One system of record.
Agent Passport
Every agent gets a signed, vendor-neutral identity record, credentials, scopes, owner, tier, lifecycle state. The same passport is planned for Microsoft agents and the agents you build, including custom Python and BYOA.
Policy Envelope
Declarative boundaries that travel with the agent. Approval routing, data scopes, blast-radius caps, dual-control for high-risk actions, enforced at runtime, not by checklist.
Audit Replay
Every decision an agent makes is recorded as a signed trace, inputs, prompts, tools, outputs, approvals. Replay any run, export to your evidence system, satisfy AI Act Article 12.
Cost Attribution
Spend rolled up by agent, model, owner, project, and business unit. Token, compute, and tool cost in one ledger, no more reconciling vendor invoices by hand.
Portable Runtime
The architecture decouples the control plane from the runtime, so the same passport, envelope, and evidence pipeline travel with the agent across the connections Meridian supports: Microsoft agents and customer-built agents at launch.
The compliance posture your auditors expect.
We have not completed a SOC 2 attestation. We will pursue an attestation when a customer requires it.
Financial Services
- SOC 2 Trust Services Criteria alignment
- FFIEC guidance
- EU AI Act high-risk
- Model Risk (SR 11-7)
Healthcare
- HIPAA controls mapped
- SOC 2 Trust Services Criteria alignment
- EU AI Act high-risk crosswalk
Professional Services
- SOC 2 Trust Services Criteria alignment
- ABA Model Rule 1.6 alignment
- EU AI Act
- Client matter scoping
Priced on agents under management.
Agents are the unit of value and the unit of risk, so that is what we price on. Not seats, not tokens. Four tiers that scale with your workforce.
For first deployments and pilot scopes.
- Unlimited Workforce Seats
- 5 Control Seats
- 25 governed agents
- 3 environment connectors
- Agent Passport, Policy Envelope, Audit Replay
- Email + Slack support
Single business unit going to production.
- Everything in Starter
- 15 Control Seats
- 100 governed agents
- 8 environment connectors
- Cost Attribution dashboards
- Priority support · 8×5
Multi-BU rollout with regulatory load.
- Everything in Pro
- 40 Control Seats
- 500 governed agents
- 20 environment connectors
- Portable Runtime · custom retention · HIPAA controls-mapped patterns
- 24×7 support · named CSM
Org-wide governance and bespoke terms.
- Everything in Business
- Unlimited Control Seats & agents
- Unlimited connectors
- Customer-managed keys through AWS KMS (BYOK)
- HIPAA controls-mapped patterns · BAA available on Enterprise contracts
- Deployment model scoped per engagement
- Custom retention and SLAs
- Dedicated TAM + solution engineering