Models you choose
Power each agent in AAES Operate with your own model access.
OpenAI
Anthropic
Google GeminixAI Grok
Azure Foundry
- Self-hosted
Self-hosted models such as GLM, Kimi and Qwen use an OpenAI-compatible endpoint.
The operating and governance layer for enterprise AI agents.
Export customer records
Share a copy of 2,400 customer records with a registered recipient through the example capability crm.contacts.export.
Approval applies to this request only, with approvers chosen by your policy.
| Agent | Committed | Ceiling | Utilization |
|---|---|---|---|
| Reporting Agent | $10.00 | $50.00 | |
| Support Agent | $0.00 | $20.00 | |
| Data Agent | $0.00 | $45.00 | |
| Release Agent | $0.00 | $25.00 |
Illustrative aggregation of task-budget reservations by agent. This display is not a vendor billing meter; reservations count against configured ceilings and do not measure or cap final vendor charges.
Reporting Agentcrm.contacts.export (example; share) · 2,400 recordsPriya RamanAuthorized#4181 · 77b1c0e2…03aa9ab42a…0c91Two product examples, shown in one frame.
AAES Operate
Agents spread across teams and tools, with no shared view of who owns them, what they work on, or what they cost.
48%
say their AI agents operate in silos, not woven into end-to-end business processes.
Camunda, State of Agentic Orchestration and Automation, 202694%
of IT leaders are concerned about AI sprawl. Only 12% use a centralized platform to manage it.
OutSystems, State of AI Development, 2026AAES Govern
An agent may cross several applications, each with its own permissions. You need to know which actions are checked before they run, where a bypass remains, and what the record keeps.
80%
of companies say their AI agents have taken unintended actions, including accessing unauthorized systems (39%).
SailPoint, AI agents: The new attack surface, 202574%
say agents often receive more access than necessary. 68% can’t clearly distinguish human from AI agent activity.
Cloud Security Alliance, Identity and Access Gaps in the Age of Autonomous AI, 2026AAES Operate organizes and runs your agents. AAES Govern checks and records the actions routed through it.
Give the agent a role and a manager.
Assign the task and follow its progress.
Check routed requests first.
Keep a decision record you can verify.
Combined workflow: routing is evaluated per workflow during the design partnership. Govern checks only the actions routed through it.
Scope your workflowCreate agents with the models you choose, or connect the agents you already run.
Power each agent in AAES Operate with your own model access.
Self-hosted models such as GLM, Kimi and Qwen use an OpenAI-compatible endpoint.
Keep the agent and runtime you use today.
Connect your own agents to AAES Operate through a webhook or command.
Organize people and agents in one chart, run projects and scheduled work, and equip agents with skills and connectors.
Checks permissions, approvals, and reserved spend on routed actions, then signs a record you can verify offline.
Capability library & packages Packages define the actions an agent can request. Connectors provide the routes used to carry them out.
AAES Govern can block an action when it controls the required credentials and the agent cannot bypass that path.
At Meridian, Reporting Agent prepares customer reporting. AAES Operate gives the agent a role, tools, and a task to own.
Explore AAES OperateGive agents roles, managers, and visible reporting lines.
Assign tasks, choose models, attach skills and tools, and schedule recurring work.
Require approvals, grant tools, and stop work when recorded spending reaches a limit.
Customer reporting · Normal priority
2,400 records · Request APR 2410
Customer reporting · Scope confirmed
Operate keeps the work on track: approvals between steps, tool access per agent, and budgets that stop work after recorded spend reaches a limit.
Checking each action before it runs, with spending reserved first and a signed record you can verify offline, is AAES Govern.
On requests routed through AAES Govern, it checks permissions and required approvals, and reserves the evaluated amount before authorization. It can block the request when it controls the credentials and the agent cannot bypass that path. It records the decision.
The agent names the capability it wants to use. On the enforced path, the destination and credential come from the registration, never from the agent.
Reporting Agentcrm.contacts.export2,400 recordsThese controls apply to requests routed through Govern; observation-only integrations record reported activity without blocking it.
Spending is reserved before authorization. This limits commitments, not final provider charges, model usage, or cloud costs.
The export decision becomes record #4182, shown below.
Check its integrity without connecting to an AAES service.
Reporting Agentcrm.contacts.export (example; share) · 2,400 recordsPriya Ramanallow_share · R3$10.00 / $50.00Enforcede94d…f7c077b1…03aa3f9c…8e21Verification needs a separately trusted public key. It checks export integrity, not complete capture or downstream execution. Stored authorization, scope, and approver fields also need the records sidecar and a separate binding check. See what the public verifier checks →
Keep your identity provider, collaboration tools, vault, and monitoring. AAES Govern connects them to agent permissions, approvals, and action records. AAES Operate connects agents to your tools through its own connector library.
Identity & directory
AAES Govern operator sign-in via OIDC. Not an agent action path; without an IdP configured, operators use a client-set bearer token.
Identity & directory
AAES Govern operator sign-in via OIDC. Not an agent action path; without an IdP configured, operators use a client-set bearer token.
Directory & collaboration
Directory and surface registration. An agent holding its own workspace credential bypasses AAES Govern; that path is observational at best.
Enterprise communication
Capability pack, enforced: the bot token stays in your vault and AAES Govern injects it per call. Bypass: a token issued directly to the agent.
Secrets & key custody
The enforced path's credential source: AAES Govern holds a reference and fetches at use. Bypass: material distributed outside the referenced path.
Evidence & monitoring
AAES Govern record export for your monitoring. Observational: it receives evidence and prevents nothing.
Tested against simulated vendor APIs. Compatibility with your live environment is validated during the design partnership.
Tell us which tools your agents need to use. We’ll confirm what each product supports, explain any limitations, and define what needs validation in your environment.
Both products are self-hosted. You operate the infrastructure, keys, storage, and availability.
Deployment modes are designed to be shared; only AAES Govern’s are verified today. Operate deployment is scoped in the design partnership.
Run AAES Govern with Docker Compose. Records live on your host. You create and control the keys.
Reference deployment configuration in the design partnership.
You operate keys, storage, and availability.
Bring one real workflow, a named operator, and weekly feedback.
We scope AAES Operate, AAES Govern, or both with you before anything is signed.
Test your live environment and document coverage and gaps.
Agree on product scope and acceptance criteria for your workflow. See the proposed outputs for AAES Govern and AAES Operate; the signed agreement fixes them.
USD 40,000 for three months, covering both products and starting after signing and onboarding. No license fee per person or per agent.
Where we stand today: no SOC 2 report exists, and there are no design partners or customer evaluation results yet.
Read the full status in the trust center →
AAES (Autonomous Agentic Enterprise Systems) makes two self-hosted products for enterprise AI agents. AAES Govern checks whether an agent may take a requested action: for actions routed through it, it checks permissions, required approvals, and spending limits, then records the decision. AAES Operate runs and coordinates agents on your infrastructure, with its own approvals, tool permissions, and budgets; it does not include Govern’s controls. Use either one, or both.
With Govern, each registered agent has at least one accountable manager, ultimately answerable to a person, and employee access stays in your existing identity provider. On enforced paths, a request needing a new decision cannot proceed if Govern cannot record that decision. Previously issued grants can remain usable for up to 15 minutes from issue. Govern cannot stop observation-only activity or calls that bypass its controlled path. You operate AAES on your infrastructure; no availability SLA is offered today.
If you already run agents and need control over what they do, start with AAES Govern: connect the agents you run and route their actions through it. If you need to create, run, and coordinate agents, deployed on your infrastructure and supported by AAES, choose AAES Operate. Whether the actions of agents that Operate runs can be routed through Govern is evaluated in the design partnership.
Not in the AAES Govern sense. AAES Operate has its own operational controls: approval steps, per-tool permissions, and budgets that stop work once recorded spend reaches a limit, so work already running can overrun. It does not include AAES Govern. Govern adds spending reserved before an action is authorized, signed decision records you can check offline, and governance for agents you already run on other platforms. Govern applies only to actions routed through it.
If AAES Govern is unavailable or cannot write the required decision record, requests needing a new decision are refused on enforced paths. Govern authorizes no new grant or brokered execution through that path. This behavior is called fail closed.
Previously issued grants remain usable until expiry, at most 15 minutes from issue. An outage does not cancel work already authorized, undo downstream effects, or stop observation-only and bypass activity. Employee access stays with your identity provider. You operate AAES on your infrastructure; no availability SLA is offered today.
The design-partner offer is USD 40,000 for three months and covers AAES Govern and AAES Operate; the term begins once the agreement is signed and onboarding is complete. The platform evaluation package for AAES Govern and AAES Operate runs in your environment and is included only in the paid design partnership.
Production rates are unpublished. Neither product has a license fee per person or per agent, and neither has a licensed-agent-count limit. For AAES Govern, the planned production model meters governed-decision capacity: capacity for requests Govern evaluates and records, including permits and refusals. The planned production model for AAES Operate is an annual subscription per production deployment, including high-availability nodes, non-production environments, setup, and standard support. Other charges and billing details remain unsettled; the pricing comparison explains those open questions.
If you stop using AAES Govern, you retain your exported records and the offline verifier supplied with your deployment. Check exported AAES Govern records for integrity without connecting to an AAES service. Offline verification checks export integrity using a separately trusted public key. It cannot prove complete capture or downstream execution. An optional independent witness or timestamp authority can corroborate the signed journal summary.
The platform evaluation package for AAES Govern and AAES Operate runs in your environment and is included only in the paid design partnership.