AEGIS is CompWarden's platform for exactly this. It checks every comp against your delegation-of-authority matrix, routes it to whoever actually holds the limit, and seals the decision to an audit trail no one can alter — turning policy from a spreadsheet people are meant to remember into a service every system consults before value leaves the building.
Comps are one of the biggest discretionary spending levers in an integrated resort, and one of the least controlled. Every department can give value away: meals, suite upgrades, show tickets, limos, free play, loss rebates that reach seven figures for a single patron. This generosity is not waste. It is the reinvestment engine that keeps players loyal. But it moves cash-equivalent value through hundreds of hands, around the clock, with almost no real-time governance to match the scale of the exposure.
Most properties document comp authority in a spreadsheet, then assume the floor will follow it. Limits exist as text, not as a rule the system can enforce.
Nothing checks whether a comp is within bounds before it is issued. Small approvals accumulate, requests get structured to stay under thresholds, and favours move outside any governed process.
Patterns surface retrospectively, long after the value has left the property. Compliance ends up depending on individual judgment and goodwill, which is not governance.
The comp matrix stops being a document people are supposed to remember and becomes a service every system on property consults before value leaves the building. Any terminal, workflow, or AI agent asks the same question first: is this authorised, right now, by someone who holds the limit?
Any system can automate a comp. Only a governed, audited, cryptographically accountable platform can automate one in a way that regulators, auditors, and executives can stand behind.
The floor keeps moving. Routine comps resolve in seconds, and anything above a limit routes itself to the right person automatically, with the guesswork removed.
An F&B manager comps a meal. If it sits inside her authority with the required evidence attached, it approves on the spot.
The request climbs the reporting line until it reaches someone whose authority covers it. At 3am it routes to the duty manager's mobile app, not a day-shift phone.
Approvals happen in a few taps. Delegation covers leave, and a licensing registry ensures authority is only exercised by credentialled staff.
The authority matrix, resolved at runtime
| Role | Limit | On exceed |
|---|---|---|
| Casino Host | $5,000 | Escalate |
| VP Player Development | $10,000 | Escalate |
| SVP Marketing | $50,000 | Escalate |
| IR President | $1,000,000 | Hold for review |
Resolved against a live org chart across every department and property, not a static spreadsheet.
The same governed record serves the floor, Finance, audit, and the regulator, each getting what they have always needed and never had in real time.
No more hunting for a signature while a patron waits. Routine comps resolve in seconds and escalations route themselves, so hosting stays generous without slowing down.
A live view of every dollar of comp value leaving the property, by venue, approver, patron tier, and comp type. Track spend against budget as it happens, and model a matrix change against months of real history before you publish it.
Audit teams spend most of their time collecting data and little of it analysing. AEGIS flips that. Records are already there, already linked, already immutable, so audit shows up to ask why, not to excavate what.
A self-serve portal lets regulators query the immutable ledger directly by date, patron tier, comp type, or approver. The operator configures exactly what the portal exposes, and a single button press proves the record has not been altered.
An agent watching the floor through a real-time pattern layer can offer a small comp within its own delegated limit in seconds, with no human involved. Go beyond that limit and the request becomes a normal item in a person's inbox. The agent gets no special override, and executives can trust it precisely because of that.
The comp is issued instantly and automatically. The agent acts as a governed principal in the matrix, with no special access and no exceptions.
The request routes to a human, exactly like any employee over their limit. Same rules, same ledger, same proof.
Audit integrity is layered twice, so the record holds up even against someone with full system access.
Every event is chained inside the operational database. Any alteration breaks the chain and shows.
Events are anchored into immudb, an open-source cryptographically verifiable store that the platform itself cannot alter.
Every sealed decision carries its ledger anchor and the exact matrix version in force when it was made:
ledger 0x9f2a…c41b matrix v4.2 sealed 2026-08-09T14:02:11ZThe matrix is versioned data, not code. That is what makes deterministic replay and honest what-if simulation possible at all.
Replay any historical decision against the exact rules in force at the time. Reconstruct the organisation as it existed on any date. Simulate a proposed change against real historical volume before publishing it.
Two engines, deliberately separate: a decision engine that answers authority checks in milliseconds for a terminal on the floor, and a workflow engine that runs escalations, reminders, and holds without blocking real-time responses.
Containerised, stateless services with active-active deployment across data centres. A sanctioned offline path keeps the floor operating if connectivity drops, reconciling automatically on reconnect.
No vendor lock-in. Adding a second property is a configuration change, not a new implementation project. Deploy on-premises, cloud, or hybrid with consistent behaviour.
Every decision is broadcast the moment it happens on an Apache Kafka backbone, so analytics and reporting feeds consume it live without slowing the approval path.
Every AI agent call is a governed transaction: logged, timestamped, and tied to a delegated limit. Agents from any vendor connect through the same protocol, with no side doors.
Query the authority engine, inspect matrix versions, replay decisions, and manage agent registrations from the terminal. Infrastructure teams automate governance like everything else.
Exclusion checks run before any incentive reaches a patron, and responsible-gaming caps and cooldowns are set per comp type, so a gesture of hospitality never becomes an inducement.
The problem sits in a blind spot between three industries. Corporate DOA tools record who should have authority for auditors, but cannot answer a terminal in milliseconds. Approval-workflow products understand invoices, not patrons or theoretical win. Casino management systems know the floor, but only their slice of it. Nobody built the thing in the middle.
Understands theoretical win, patron tiers, exclusion lists, and gaming-specific comp types. No incumbent does.
Resolves authority against a live org chart across every department and property, not a static file.
Every agent operates under the same matrix as a human employee, with its own delegated limits and full audit trail.
Every decision is signed and chained. No record can be altered after the fact, by anyone.
AEGIS is looking for two kinds of partners at this stage: operators who want to be first to market with governed comp automation, and investors backing infrastructure at the intersection of AI, compliance, and gaming.
For integrated-resort operators who want to shape the roadmap, secure preferential commercial terms, and be first to market with a capability their competitors do not have.
We are raising to accelerate product development, expand the engineering team, and establish the first enterprise deployments. If you back infrastructure plays at the intersection of AI, compliance, and gaming, we would like to talk.