Let AI agents work.
Keep your staff
in control.

The operating and governance layer
for your AI workforce

The problem

Most companies use AI.
Most don’t report financial gains.

Nearly nine in ten respondents report regular AI use in at least one business function.Only 37% report a positive contribution to earnings (EBIT). McKinsey, The state of AI in 2026: On the road to ROI, 2026Nearly nine in ten respondents report regular AI usein at least one business function. Only 37% reporta positive contribution to earnings (EBIT). McKinsey,
The state of AI in 2026: On the road to ROI, 2026

To deliver value, an agent needs five things.

  1. Structure

    Projects and tasks inside the company, with the skills and app connections the work needs.

  2. Ownership

    A named person responsible for the agent.

  3. Access

    The data and systems its work requires.

  4. Guardrails

    Action and spending limits, with approvals where needed.

  5. Audit trail

    A record of its actions that you can check.

Agents work outside the organization

AAES Operate

When an agent’s work sits outside shared projects and tasks, it’s hard to say who is responsible for it.

84%

of surveyed companies have not redesigned jobs around AI capabilities.

Deloitte, State of AI in the Enterprise, 2026

50%

of AI agents operate in isolated silos, according to surveyed IT leaders.

Salesforce / MuleSoft, Connectivity Benchmark Report, 2026

Access is approved in meetings

AAES Govern

Access is approved once, up front. Agents also need limits on each action they take, and a record of what was allowed.

77%

of surveyed organizations report that AI adoption is outpacing their governance capabilities.

IBM, The 2026 Tech Leader Study, 2026

47%

of surveyed leaders at large US public companies say their organization skipped its AI governance process for urgent deployments.

EY, US AI Risk and Governance Survey, 2026

The solution

AI employees,
managed by your staff.

Give AI employees defined roles, tools for their work
and accountable human managers.

Agents

Org chart · people and agents
6 people7 agents3 teams
  • DWDana WhitfieldChief Operating Officer
    • PRPriya RamanHead of Data
      • RAReporting AgentCustomer reportingAnthropicRunning
      • DAData AgentData retentionOpenAIRunning
      • RSResearch AgentMarket researchGoogle GeminiIdle
      • LOLena OrtizData Analyst
    • MCMaya ChenHead of Client Ops
      • SASupport AgentClient follow-upsOpenAIRunning
      • RNRenewal AgentRenewal briefsGoogle GeminiPaused
      • SOSam OkaforSupport Lead
    • JLJordan LeeHead of Engineering
      • RLRelease AgentService deploymentsAnthropicRunning
      • QAQA AgentRelease checksxAI GrokIdle

Example screens from Operate and Govern.

How it works

Power agents with AAES Infer.
Build, equip and run agents with AAES Operate.
Govern their actions with AAES Govern.

Choose where your AI models run, give agents work in projects and tasks,
and approve and record the actions they send through Govern.

Step 1 Set up your organization

Import your organization

Bring in your staff and existing reporting lines.

  • Microsoft Entra ID
  • Okta
  • Google Workspace
or

Create your organization manually

Add your staff, create teams and define reporting lines.

StaffTeamsReporting lines

Step 2 Add your agents

Self-hosted or AAES-hosted

Create a new agent in AAES Operate

Operate runs your agents and assigns their work.
Govern checks the actions they send through it.

  • OpenAI
  • Anthropic
  • Google Gemini
  • xAI Grok
  • Azure AI Foundry
  • Self-hosted
Also through AAES Govern
or
Runs on your platform

Connect an existing agent

Keep your agents where they already run.
Send their actions through Govern.

  • Copilot Studio
  • Azure AI Foundry
  • Your runtime
AAES Infer

Model access for the agents you run

Self-hosted or AAES-hosted
  • Location

    Your chosen jurisdiction

    Requests outside it are refused

  • Spending

    Company and project limits

    A shared prepaid balance or your provider account

  • Data use

    No training on your data

    Prompts and outputs are not reused

Explore AAES Infer

Step 3 Govern their actions

AAES Govern
Self-hosted or AAES-hosted
  1. Who

    Each agent answers to a person

    Limited permissions

  2. Check

    Every request

    Against your rules, approvals and limits

  3. Act and record

    In your connected apps and tools

    A record of each decision and action

Works with your apps, tools and data
Power AI employees / AAES Infer

Choose where
your models run.

Use AI models that process your requests in the country or region you choose. You get one company bill, and provider terms that forbid training on or reusing what you send and receive. We confirm each offer for your case before it starts.

Explore AAES Infer
Choose where your AI requests are processedAAES Infer is for LLM processing in the jurisdiction you choose, with no training or secondary reuse of prompts and outputs. The illustrated jurisdictions are Australia, Hong Kong, Japan, Singapore, Switzerland, the United Arab Emirates, the United Kingdom, the United States and the European Union. Each model and location requires verified provider terms and controls before activation. Pins identify geographic areas, not individual data centers.
Scroll the map to see all jurisdictions.
Operate your AI workforce / AAES Operate

Give your AI employees tools.
Track their progress.

With Govern’s job-profile tools, configure the tools each AI employee needs
inside your company’s own Microsoft 365, Google Workspace or cloud account.

Mailbox

Its own email address

Chat

Slack or Teams handle

Browser

Isolated web browser

Payment card

Virtual payment card

Phone number

Its own phone number

Desktop

Virtual desktop for legacy apps

Workspace provisioning uses AAES Govern’s job-profile tools and requires configured provider accounts, permissions and any required licences. It is not an automatic step when hiring an agent in standalone Operate. Vendor adapters have been tested against local fixtures; Microsoft 365 and Google Workspace live-tenant qualification remains outstanding. Confirm provisioning and retirement in your own environment before relying on them.

Track its work.

Each AI employee gets a role, a manager and its own tasks. You see its progress on the board, and the decisions it needs go to its manager’s inbox.

  • Role
  • Manager
  • Tasks
  • Decision inbox
AAES Operate
Work / Projects

Meridian · Customer reporting

Task board
To do1
  • CRM-25Prepare renewal brief

    Customer reporting · Normal priority

    Customer Ops
In progress1
Done1
  • CRM-23Confirm export scope

    Customer reporting · Scope confirmed

    Reporting Agent
Follow Reporting Agent’s export request into the checks below.

On its own, Operate starts no new work once recorded spending reaches the budget, and work already running can go over. With AAES Govern, money for every model call and action sent through Govern is set aside from the budget before it runs.

Explore AAES Operate
Govern AI employee actions / AAES Govern

Follow an agent action
from request to record.

Govern checks each request, asks a person when policy requires it,
and records the decision.

MERIDIAN · APR 241001 / 06

The agent requests an action.

The agent asks to use one of its registered actions, here a customer-records export. When Govern enforces the action, the agent never chooses where the data goes and never holds the key to the system. Both come from Govern.

ActorReporting Agent
Capability (example name)crm.contacts.export
Requested data2,400 records

Govern controls only the actions sent through it. You can check its records offline.

Explore AAES Govern
Deploy in your environment / AAES Operate + Govern

Run AAES on your servers,
or let us host it.

Keep the tools you already use for staff sign-in, stored keys and monitoring.
During the evaluation, we test the connections you need.

Microsoft Entra

Staff sign-in and directory

Okta

Staff sign-in and directory

Google Workspace

Directory & collaboration

Slack

Team chat

HashiCorp Vault

Stored keys and passwords

Splunk

Logs and monitoring

We’ve tested these against simulated versions of each service. We confirm they work with your accounts during the evaluation.

Check your stack.

Tell us which tools your AI employees need. We’ll confirm support and any gaps.

Book a demo

Your boundary. Your control.

Run Operate and Govern on your own servers, or let AAES host them for you.

SINGLE VMCLIENT OPERATED

Start on one server.

Install both products on one machine you run.

Evaluation questions / AAES

Before you evaluate.

Have something else in mind?
Talk to us

What is AAES, exactly?

AAES (Autonomous Agentic Enterprise Systems) makes three products for companies that use AI agents. AAES Operate is where you create agents, give them work and track it. AAES Govern checks each action an agent sends through it against your permissions, approvals and spending limits, and records the decision. AAES Infer gives your agents AI models that process requests in the country or region you choose, on one company bill. Infer isn’t open for general sign-up. You can use Operate and Govern separately or together, on your own servers or hosted by AAES.

Govern controls only the actions sent through it. It can’t stop calls that go around it, and actions you set to watch only are recorded but not blocked. We don’t offer an uptime guarantee (SLA) yet.

Which product do we need?

Choose AAES Operate to create agents and give them work. Choose AAES Govern if you already run agents and need control over what they do. Ask about AAES Infer if your AI requests must be processed in a specific country or region, on one company bill. If you use both Operate and Govern, Operate’s agents send their tool calls through Govern, Claude and Codex agents send their model calls there too, and managers answer approvals from their Operate inbox. Live deployment qualification of AAES Operate + Govern is still in progress. Confirm the configured controls in your own deployment before you rely on them.

Does AAES Operate govern agent actions?

Not the way Govern does. Operate has its own controls: who can see private work, task reviews, and budgets that stop new work once recorded spending reaches the limit. Govern adds three things. It sets money aside before it allows an action, it keeps signed records you can check offline, and it also covers agents on other platforms.

What happens if AAES Govern is unavailable?

If Govern goes down, or can’t save its record of a decision, requests that need a new decision are refused wherever Govern enforces. This is called fail closed.

Permissions Govern already issued stay valid until they expire, at most 15 minutes after issue. An outage doesn’t undo work that was already approved, and it doesn’t stop actions that are watch-only or that don’t go through Govern.

What does AAES cost?

Pricing and evaluation terms are detailed on the product pages. Contact us to discuss your workflow. Infer pricing is not published.

Bring one job you want AI agents
to do or to control.

Try Operate, Govern or both on that one job, on your own servers.