- An operator's stack has about ten layers: PAM, game content, sportsbook, payments, KYC and AML, bonus engine, CRM, helpdesk, data and BI, and risk and RG tooling.
- The PAM is the system of record for players, wallets and transactions; almost every other layer either reads from it or writes to it, which makes its API the single most important thing to check.
- Turnkey, white-label and self-built set-ups differ mainly in who holds the licence, who owns the data and how much of the stack you can replace without the provider's permission.
- Ask platform providers about API coverage, event streams, data export, rate limits and who can build on top — not only about game count and payment methods.
- AI player support sits on top of the PAM and the helpdesk: it reads account state, writes through validated actions and hands off in the tool agents already use. It should not require replacing either.
Ask five people what iGaming software is and you will get five answers: a casino platform, a game studio, a sportsbook, a payments gateway, a CRM. All of them are right, because an online casino or sportsbook is not one system. It is a stack of ten or so layers, usually bought from several B2B iGaming providers and held together by APIs, and the way those layers are divided decides how fast you can launch a market, change a supplier or add something new on top.
This article maps the stack from an operator's point of view: what each layer owns, how turnkey, white-label and self-built set-ups differ, what to ask an iGaming platform provider before you sign, and where AI player support plugs in. It names no vendors; the point is to understand the shape of the stack, not to rank suppliers.
The layers of an operator's stack
Every licensed operator runs some version of the following. Some layers come bundled with the platform; others are almost always separate contracts.
| Layer | What it owns | Typically sourced |
|---|---|---|
| PAM (player account management) | Accounts, wallets, transactions, limits, verification status, back office | Platform provider, or built in-house by large groups |
| Game content and aggregation | Casino and live games, remote game servers, game rounds | Studios via one or more aggregators |
| Sportsbook | Markets, odds, bet placement, settlement, trading | Sportsbook provider, sometimes bundled with the platform |
| Payments | Deposits, withdrawals, payment methods, routing between processors | Payment service providers and an orchestration layer |
| KYC and AML | Identity and document checks, sanctions and PEP screening, source of funds | Specialist verification and screening vendors |
| Bonus engine | Promotions, wagering rules, free spins, free bets | Platform, or a separate gamification tool |
| CRM and marketing automation | Segments, campaigns, messaging across email, SMS, push | Specialist iGaming CRM or a general one |
| Helpdesk and telephony | Chats, emails, calls, tickets, agent workspace | General customer-service software |
| Data and BI | Event stream, warehouse, reporting, regulator reports | In-house on a cloud data platform, plus platform reports |
| Risk, fraud and RG tooling | Fraud and bonus-abuse detection, responsible-gambling monitoring | Platform features, specialist vendors or in-house models |
Affiliate tracking, content management for the front end and game-level reporting are often added as further layers. The exact split varies, but the pattern does not: one system of record and many systems around it.
The PAM is the centre of gravity
The PAM holds the facts every other layer needs: who the player is, what their balance is, which bonuses are active, what they have deposited and withdrawn, whether they are verified, which limits or exclusions apply. Games and sportsbook settle against its wallet. Payments update it. The CRM segments on its data. Support agents look things up in its back office dozens of times an hour.
That makes the PAM's API and event stream the most important thing to understand in the whole stack. A PAM with good read APIs and a real-time event feed lets you add a CRM, a fraud model or an AI support agent in weeks. A PAM whose data is only reachable through its own back-office screens and a nightly export makes every addition a negotiation with the provider. Game count and payment-method lists matter for players; API coverage matters for everything you will want to do next.
Turnkey, white label or self-built
How the stack is assembled is usually described in three models. Terms vary between providers, so treat these as shapes rather than definitions:
| White label | Turnkey | Self-built / modular | |
|---|---|---|---|
| Licence | Provider's | Operator's own | Operator's own |
| Time to launch | Fastest | Moderate | Slowest |
| Data ownership and access | Limited; check contract | Better; depends on APIs and export | Full |
| Replacing a layer | Rarely possible | Possible where the provider allows integrations | Your decision |
| Adding third-party tools | Provider's list | Via APIs, sometimes with approval | Any |
| Engineering effort | Lowest | Integration team | Platform team |
Many operators move along this table over time: start white label to test a market, move to turnkey with their own licence, and replace individual layers — usually CRM, payments orchestration, data and support — as they grow.
What to ask an iGaming platform provider
Sales conversations with platform providers tend to focus on content, markets and payment methods. Those matter, but the questions that decide your freedom later are about data and integration:
- API coverage. Which player, wallet, bonus, payment, verification and limit data can be read through the API? Which actions can be performed, and with what permissions?
- Event stream. Is there a real-time feed of deposits, bets, withdrawals, bonus events and status changes, or only batch exports?
- Data export and exit. In what format can you take your full player and transaction history if you leave, and how long does it take?
- Rate limits and sandbox. Can a third-party tool call the API at support-desk volumes? Is there a sandbox with realistic test data?
- Building on top. May your own engineers or third-party vendors integrate with the platform, and does that need the provider's approval or fees?
- Audit logs. Which back-office actions are logged, with which user and reason, in a form your compliance team and regulator can use?
- Regulatory reporting. Which markets' reports are supported out of the box, and who maintains them when rules change?
Ask the provider to show, in the sandbox, an API call that returns a single player's pending withdrawal with its status reason, active bonus with wagering progress, and verification status. If that takes one call, most integrations will be easy. If it takes a support ticket, plan accordingly.
How data moves: one withdrawal across the stack
Following one everyday event shows why the layers have to talk to each other. A player requests a withdrawal:
- The PAM checks balance, active bonus and wagering, open bets and verification status, and creates the request.
- Risk and AML rules may hold it for review; KYC may require a document first.
- Payments route it to a processor; the processor's status codes come back to the PAM.
- The CRM may suppress promotions or send a confirmation.
- If anything stalls, the player contacts the helpdesk, where an agent — or an AI agent — needs to see all of the above to give a specific answer.
- The data layer records the whole sequence for reporting and for the regulator.
Most support pain in iGaming comes from step five: the agent sees the ticket in one tool and the account in four others. The anatomy of this exact contact is in how to automate withdrawal and KYC questions.
Where AI player support fits
AI support is often sold as if it were a new platform. It should not be one. In a well-designed stack it is a thin layer between the systems you already have:
- Reads from the PAM, payments and KYC to see the player's actual situation before answering.
- Writes only through validated actions. The model proposes; deterministic rules decide whether an account change is allowed, and anything near money stays behind thresholds or with people.
- Works inside the helpdesk your agents already use, so handoffs land in the same queue with the context attached.
- Hands sensitive cases to humans by rule: responsible gambling, VIPs, disputes, fraud and AML.
- Runs in your infrastructure, with player identifiers tokenised before any text reaches an external model, so account data and audit logs stay where your other systems are.
That is how PlayerDesk is built: it reads from the PAM, payments and KYC systems and writes into the helpdesk, without replacing the PAM, the CRM or the helpdesk. Which PAMs, helpdesks, telephony, payment and KYC systems it connects to, and what happens when one of them is down, is on the engineering page. The broader case for this kind of agent is in AI player support for iGaming operators.
What PlayerDesk touches, and what it doesn't.
Deployment footprint, adapters for PAM, helpdesk, voice, payments and KYC, failure behaviour and who can switch it off — written to be forwarded.
Buy, build or add on top
For each layer, the realistic options are to take what the platform offers, buy a specialist product, or build it. A rough guide:
- Take from the platform where the layer is commodity and tightly bound to the wallet: game aggregation, basic bonus mechanics, standard regulatory reports.
- Buy a specialist where the layer is a discipline of its own and changes often: KYC and AML screening, payments orchestration, CRM, helpdesk.
- Build or own the logic where the layer encodes how you run your business: your data warehouse, your risk rules, your escalation policies, your QA rubric.
AI support straddles the last two. The engine is worth buying — building a production-grade, audited agent in-house typically takes far longer than integrating one — but the rules, knowledge base, escalation list and evaluation set describe your business and should belong to you whichever vendor runs them.
Pick platforms for their APIs as much as for their features. Everything you add later depends on them.
The short version
iGaming software is a stack, not a product: a PAM at the centre, with game content, sportsbook, payments, KYC and AML, bonuses, CRM, helpdesk, data and risk tooling around it. The PAM's API and event stream decide how easily anything else can be added or replaced, so check them as carefully as content and markets. White label, turnkey and self-built differ mainly in licence, data ownership and replaceability. And AI player support belongs on top of the PAM and inside the helpdesk, not in place of either. If you are planning that layer, how to automate iGaming customer support sets out the integration and rollout step by step.
Frequently asked questions
What is iGaming software?
iGaming software is the set of systems an online casino or sportsbook runs on: the player account management platform (PAM) that holds players, wallets and transactions, game content and sportsbook engines, payment and KYC integrations, bonus and CRM tools, the helpdesk, data and reporting, and fraud and responsible-gambling tooling. Operators usually buy most of it from several B2B providers.
What is a PAM in iGaming?
A PAM, or player account management system, is the core platform that stores player accounts, balances and transactions, manages registration, limits and verification status, and connects games, payments and back-office tools. It is the system of record that most other iGaming software reads from or writes to.
What is the difference between turnkey and white-label iGaming platforms?
In a white-label set-up the operator runs a brand on the provider's licence and platform, with limited control over integrations and data. In a turnkey set-up the operator holds its own licence and runs a complete platform supplied by a provider, with more control. A self-built or modular set-up combines a PAM with separately chosen components. Terms vary between providers, so check licence, data ownership and integration rights in the contract.
What should I ask an iGaming platform provider?
Ask which player, wallet, bonus, payment and verification data is available through the API, whether there is a real-time event stream, how data can be exported if you leave, what rate limits and sandbox environments exist, whether third parties can build on the platform, and what audit logs it keeps for regulators.
Where does AI customer support fit in the iGaming stack?
On top of the PAM and the helpdesk. An AI support agent reads account, payment and verification state from the PAM, answers in the helpdesk the team already uses, executes only validated actions and hands sensitive cases to people. It should not require replacing the PAM, CRM or helpdesk.
Last reviewed 3 November 2026. Figures marked as industry ranges are ranges, not Arctura claims; operator-specific numbers are only quoted after validation on the operator's own baseline.


