The internal AI that knows your company and acts under governance
An AI copilot for your employees: it answers with your organization's context, proposes the next step and only writes, sends or executes after an approval click on screen — with an audit trail and per-organization access.
Human approval by risk · No click, the request expires · Audit trail · Per-organization access
Prepare the quarterly summary and notify the client through Axis.
Summary ready with the company context. The client notice is awaiting your approval.
Generic AI doesn't know your company — and leaves no trace
Knowledge and decisions scattered. An AI that doesn't understand the business context, has no clear limits and acts without leaving proof of what it did. For a company, that's risk, not productivity.
No context
Generic AI answers into the void: it doesn't know your organization's clients, processes or decisions.
No limits
Without authority levels, any action seems allowed. What should require approval slips through.
No trace
When no one can say who asked, what ran and when, there is no way to audit — or to trust.
An AI that belongs to the company — and answers for it
Axiom works with your organization's context, proposes the next step and only writes, sends or executes after a person approves on screen. Generic module reads and writes are recorded in your organization's audit trail.
With the company context
Answers and tasks start from your organization's knowledge, not from a generic AI.
With risk-based governance
Lookups run directly; anything that writes, sends or executes a plan waits for a person's explicit approval, with the risk flagged on the card.
With a record of what ran
Action cards keep what was proposed, approved and executed, and for generic module reads and writes the audit trail notes who asked, what ran and when.
An internal copilot with a handbrake
Real support for employees' work — always within your organization's rules.
Copilot in sessions
An AI assistant that supports employees in tasks and decisions, organized into work sessions.
Actions under risk-based governance
Reads run without approval; in the copilot, every action that writes, sends or executes a plan stops at an approval card on screen, with its risk level flagged.
Recorded, auditable actions
Actions proposed, approved and executed on cards are recorded, and generic module reads and writes enter the audit trail with who asked, what ran and when.
Client notice through Axis
Axiom can send client notices through Axis after approval on screen, with recipients resolved only inside your organization.
Per-organization access
The organization of each request is resolved on the server from the verified identity, among the companies where the person is an active internal member, and generic module reads and writes follow the database access policy.
AI usage cap
AI usage has a cap: Axiom warns at 80% and blocks at 100%, with no surprises. Billing stays outside the system.
An action with governance, from request to record
The path a request takes when it becomes an action in Axiom: context, proposal, human approval on screen and execution.
From the employee's request to an approved action
IllustrativeEmployee makes a request
An employee asks a question or requests a task from Axiom, in natural language, inside a session.
Axiom answers with the company context
The assistant uses the organization's knowledge to answer and frame the request.
Proposes an action
When acting makes sense, Axiom proposes the concrete action — without executing on its own.
Human approval on screen
An action that writes, sends or executes stays locked on a card until the person decides; with no click, the request expires and is refused.
Execution after the click
Once approved, the action runs; on session cards, the proposal, the approval and the execution are recorded.
If it involves the client, the message goes out with approval
One route is Axis, with recipients scoped to your organization; nothing goes out to the client without the approval click.
Illustrative flow: the employee makes a request, Axiom answers with the company context, proposes an action, the action waits for a person's approval on screen, it is executed and, if it needs to reach the client, it can go out through Axis.
Nothing risky happens without someone approving
Human approval on screen, risk flagged on the card and an audit trail are part of the product — not an optional feature.
Approval before acting
Reading runs without approval; whatever writes, sends or executes a plan waits for a person's explicit click, with the risk level flagged on the card.
Recorded actions
Session action cards keep what was proposed, what was approved and what was executed.
Auditable trail
Generic module reads and writes are recorded in your organization's audit trail, with who asked, what ran and when.
What your team gets done with Axiom
Genuine internal support, without giving up control.
Answer with what the company already knows
An employee asks and gets an answer anchored in the organization's context, not in a generic AI.
Execute tasks with on-screen approval
Operational tasks are proposed by Axiom and only run after the person approves on the card; with no click, the request expires.
Notify the client without losing control
When the client needs to be notified, the message can go through Axis, with recipients from your organization and after your approval click.
Axiom decides inside; Axis is one of the routes to the client
The connection below was verified in the code. Axiom acts internally; notices to the end client can be delivered through Axis.
The connection above was verified in the code, but it is not the only one: with approval on screen, Axiom also acts in other enabled Centriu modules and sends e-mail from the person's own connected Google account. Each system has its own access and per-organization scope. If your operation depends on a link that isn't here, talk to us before signing up.
It comes as a project, with priority support
Axiom is rolled out as a project guided by Centriu. For agencies, the rollout is assisted.
Rollout project
Activation is closely guided: company context, client memory and connections configured with your team.
Priority attention
The account gets priority attention during adoption and operation of the system.
Assisted rollout for agencies
Agencies get an assisted rollout to bring Axiom into their own internal operation.
System plans
Three tiers that unlock more as you grow. On Max, you get access to everything.
Essentials to start
- Content and text generation
- Voice recognition in Portuguese
- Long document summarization
- Conversation history per user
Full operation
- Content and text generation
- Voice recognition in Portuguese
- Long document summarization
- Conversation history per user
- Vision and image analysis
- Table and spreadsheet analysis
- Configurable tone of voice
Everything unlocked
- Content and text generation
- Voice recognition in Portuguese
- Long document summarization
- Conversation history per user
- Vision and image analysis
- Table and spreadsheet analysis
- Configurable tone of voice
- Action execution inside the platform
- Knowledge base search (RAG)
- Integration with the contracted Centriu systems
- Proactive context-based suggestions
Frequently asked questions
What is Centriu Axiom?
It is Centriu's internal enterprise AI: a copilot for your company's employees that answers with the organization's context, proposes actions and only writes, sends or executes after an approval click on screen.
Who is Axiom for?
For the company's internal teams and employees. It is an AI that works inside your operation, with per-organization access, not a tool aimed at an external audience.
Does Axiom act on its own?
Not without control. It proposes the action, and whatever writes, sends or executes a plan stays locked on an approval card on screen; with no click, the request expires and is refused. Nothing sensitive happens without someone approving.
How do I know what the AI did?
Actions proposed, approved and executed on session cards are recorded, and generic module reads and writes enter your organization's audit trail, with who asked, what ran and when.
Does Axiom talk to my client?
Only after your approval on screen. Axis is one of the routes: the notice goes out via inbox and push, with recipients resolved only inside your organization, never to clients of another company. Axiom can also send e-mail from the person's own Google account.
How does the AI usage limit work?
AI usage has a cap: Axiom warns at 80% and blocks at 100%, with no surprises. Platform billing stays outside the system.
An internal AI your company can audit
Talk to a specialist and see Axiom applied to your team: company context, actions under on-screen approval and an audit trail of generic module reads and writes.