Agent copilot
Suggests the reply to the support rep in real time, during the conversation itself.
Apolo
For authorized operators in Brazil
Governance that runs before every decision.
Apolo is the operator's copilot: it applies caps and guardrails, classifies the risk of every decision and suggests the reply to the support rep. It never talks to the player.
Schedule a demo
The button opens a WhatsApp conversation with the Cortex sales team. No form and no sign-up.
Decision under reviewabove-average withdrawal · player #064B17A2
demo · sample dataGuardrailVerdict
Daily contact capcontacts with this player within today's cap
within
Quiet hoursoutside restricted hours · 14:32
within
Responsible gaming3rd withdrawal in 24h · Nêmesis flagged
review
Level N3 — goes to human approvala proposal with an expiry · does not execute without approvalN3
Decisions with no trail, caps with no control and alerts nobody reads turn into risk at the next audit.
Meet Apolo
The copilot that governs the operation
Caps and guardrails
Daily contact cap, quiet hours and opt-out applied before any outbound message.
Risk in four levels
Classifies every decision from N1 to N4; N3 requires human approval and N4 blocks.
Monitored anomaly
Error and guardrail deviations become classified alerts for the supervisor.
Never talks to the player.
Sending to the player is switched off: every attempt is blocked before it leaves and the block is recorded in the audit trail. Apolo works in the backoffice and in the operator's panel — the conversation with the player belongs to Atena.
Allowed destinationsblocked before the output, not in config
Outbound channelRoute
Backofficesuggestion to support rep · risk panel
allowed
Agent panelalert, anomaly and report
allowed
PlayerWebChat, email, SMS, push — any channel
blocked
Offsending to the player · every attempt recorded as a block
N3 and N4nothing with material risk goes out without a human
Contact capschecked on every decision, not by sampling
demo · sample dataContract contact cap76% of the cap
Contact cap per player48% of the cap
Messages in window3 / 3
Flow stopped at the capdefault is fail-closed — at the limit, it stopsat cap
Cap and guardrail before the output.
Daily contact cap, quiet hours, opt-out and channel health are checked on every decision. A violated guardrail ends the flow: the default is fail-closed.
High risk calls for a human.
Every decision gets a level from N1 to N4. N1 and N2 go through on their own inside the backoffice and never reach the player; N3 becomes a proposal with human approval and an expiry; N4 emits a blocking event immediately.
Risk classificationevery decision gets a level before going out
N1 · goes alonestays in the backoffice · does not go to the player
N2 · logs and goesevent written to the audit trail · also without leaving the backoffice
N3 · human approvala proposal with an expiry
this case
N4 · blocks at onceimmediate event to supervisor
Sent to supervisor in 1.4swith the supporting data attached
WORM trailevery decision becomes a record the moment it happens
recordingEventHash
Guardrail violated, decision blockedapolo.guardrail_violated · apolo
a7c30e94…
Proposal created for approvalapolo.proposal_created · apolo
2be81f57…
Output to the player blockedapolo.cap_blocked · apolo
c1049d6b…
Chained hash · PII redacted before writing · retained 5 yearsa7c30e94b52d18ff0e6a3c7d914bb02f6ce7481…
Neither the operator nor Cortex rewrites the past.
The application only writes: it never changes, never deletes. Changes and deletions are blocked in the database itself, and any break in the hash chain shows up in the check — this is what is called a WORM audit trail. Personally identifiable data (PII) is redacted before writing, the trail is kept separate per client and disposal only happens at the end of the retention period, as the LGPD requires.
Live in 15 to 30 business days
Guided rollout
No backoffice swap
15-day hypercare
Tuned to your operation
Connect your operation
We plug Apolo into the backoffice and agent panel you already run.
Define caps and authority
Daily contact cap, quiet hours and approval level become settings in your contract.
Go live with hypercare
Go-live monitored for 15 days, with anomalies and pending proposals in plain sight.
Frequently asked questions
Does Apolo talk to the player?
No. Sending to the player is switched off: every attempt is blocked before it leaves and the block is recorded in the audit trail. Apolo works in the backoffice and in the operator's panel; the one who talks to the player is Atena.
What does the copilot suggest to the support rep?
An answer anchored in the approved knowledge base, delivered during the conversation. The support rep accepts, edits or ignores it — and that feedback is recorded to calibrate the agent.
How do caps and quiet hours work?
A daily contact cap per player, night quiet hours and opt-out are checked before every outbound message. In Event Mode the cap rises in a controlled way, without releasing the other guardrails.
When does a decision require human approval?
At level N3: Apolo issues a proposal with an expiry that only executes once approved. N4 is blocking and generates an immediate event for the supervisor.
What does the record for the regulator look like?
Every governance decision enters the WORM audit trail the moment it happens, with a chained hash, PII redacted before writing and separation per client. The application only writes: changes and deletions are blocked in the database, and any break in the chain shows up in the check. The evidence comes out ready for audit — filing with the SPA/MF (Secretariat of Prizes and Betting of the Ministry of Finance) is up to you.
How much does it cost and how is it billed?
1 credit is 1 full Atena answer; an instrumented Apolo suggestion debits a fixed weight from the same pool. The full table goes in the proposal. Each plan includes a monthly fee, a setup fee equal to the monthly fee and a monthly credit pool.
What if the credit pool runs out?
All agents pause, compliance included, until top-up. No conversation is cut mid-way: whatever started is finished, and your team keeps working from the Backoffice, because human service consumes no credit. Usage is metered per cycle and per agent and stays available for review at any time; there is no automatic usage warning, so watching the balance is up to the operator. The legal duty of responsible gaming and anti-money-laundering remains the operator's, with or without credit.