Only one in five gaming companies says it has earned a meaningful return on its AI investment, according to the State of AI in Gaming 2026 report published in April 2026 by KPMG and UNLV’s International Gaming Institute, which surveyed 83 land based and online operators and suppliers.

Cevro is the AI player support platform built for iGaming, running AI agents that resolve player queries end to end across bonuses, KYC, payments and responsible gaming.

That gap between what operators were sold and what they actually got is rarely a model problem. It’s usually a procurement problem, created by the questions nobody asked during the demo.

This guide gives you the 10 questions that reveal whether a vendor can resolve tickets or only answer them, plus a way to score what you hear.

Key Takeaways

• Only one in five gaming companies reports meaningful AI returns, per KPMG’s 2026 report.

• Ask whether the AI executes back office actions or only explains policies to players.

• Verify named PAM, payment and helpdesk integrations before you sign anything.

• Pair every automation claim with CSAT, escalation quality and full audit logs.

• Cevro resolves up to 90% of tickets with CSAT above 4.8 across 50+ markets.

Why iGaming Support Costs Are Climbing Faster Than Headcount Budgets

Support volume tracks revenue, and revenue is growing fast. SOFTSWISS reported in September 2026 that the global iGaming market will generate €307.3 billion in gross gaming revenue this year, citing H2 Gambling Capital data showing a compound annual growth rate of nearly 30% between 2019 and 2026.

Every new market adds languages, time zones and rules. It also adds payroll.

Open iGaming job listings from mid 2026 show the real numbers. A Limassol operator advertised player support roles at €18,000 to €20,500 gross per year in July 2026, while Rush Street Interactive listed US player support agents at $20.00 to $24.00 per hour.

In the United States, the all in cost of one full time support agent runs $52,000 to $78,000 a year once you include benefits, tools, training and overhead.

Per ticket math tells the same story. Live chat costs roughly $5 to $9 per contact, and Gartner puts the median assisted channel contact at $13.50.

Run that against a mid sized operator handling 100,000 chats a month and you’re looking at $500,000 to $900,000 in monthly chat cost alone. That’s why support automation stopped being a nice idea and became a board level line item.

Why Most AI Support Projects Stall Before They Reach Scale

Here’s the problem. The label “AI agent” is now on almost every product in the market, and most of them can’t act.

Gartner estimates that of the thousands of vendors claiming agentic AI capability, only about 130 actually deliver it, a pattern the firm calls agentwashing. Gartner also expects more than 40% of agentic AI projects to be cancelled by the end of 2027, mainly because of escalating costs and unclear business value.

The gaming industry is not ahead of that curve. KPMG and UNLV scored overall industry AI maturity at 45 out of 100, with online operators at 54 and land based operators at 39.

Governance is thinner still. In the same study, 58% of regulators said the gaming industry cannot effectively self regulate its use of AI, and only 13% were aware of any licensed operator with an internal responsible AI policy or framework.

For example, a generative tool connected only to your help centre will explain your withdrawal policy perfectly and still leave the player waiting. It never checked the payment service provider log.

FAQ style bots typically automate just 10% to 20% of queries, which means the expensive 80% stays with your team.

What iGaming Native Means in Practice, and Why It Changes the Shortlist

Most AI support tools arrived in iGaming by accident. They were built for retail or SaaS, proved themselves there, then picked up a gaming integration and a compliance checkbox later.

Cevro took the opposite path. It was built inside iGaming, for iGaming, from the first line of code.

That difference shows up in the detail. There’s a real gap between an AI that explains a process and one that completes it.

• A chatbot tells a player where to find their transaction history.

• A Cevro agent checks it, cross references it against the payment gateway, and resolves the issue.

The mechanism behind that is the AI Procedure (AIP) engine. An AIP is a structured, executable version of your own standard operating procedure, written in plain language so your operations team can shape it without waiting on developers. Your SOPs already exist. They’re just written for humans to read, not machines to run.

10 Questions to Ask AI Customer Support Vendors Before You Sign

Print these. Ask them in the order below, because each one makes the next answer harder to dodge.

1. Can the AI Take Action in Our Back Office, or Only Answer Questions?

This is the question that sorts the market. Ask the vendor to show a live withdrawal query, not a slide.

A resolving agent verifies identity, queries the payment gateway, cross references withdrawal limits, KYC status and responsible gaming flags, then either resolves or escalates with full context.

Without back office integration, the AI can only explain. With it, the AI can act.

2. Was the Platform Built for iGaming or Adapted From Another Industry?

Ask when the vendor’s first gaming client went live and what the product did before that. Generic platforms need to learn wagering requirements, bonus eligibility logic, player tiering and PAM behaviour from scratch, usually on your budget and your timeline.

3. Which PAMs, Payment Providers and Helpdesks Are Already Integrated?

Ask for names, not categories. Cevro lists production tested connections including EveryMatrix, SoftSwiss, Vegangster, White Hat Gaming, Playtech, Evolution, Comm100, LiveChat, Zendesk and Intercom.

If a vendor answers “we can integrate with anything,” translate that as “we have built nothing yet.” You’ll be funding the first one.

4. How Does the Platform Handle Responsible Gaming Signals?

Manual quality sampling covers roughly 2% to 7% of interactions. Ask what percentage of conversations the platform screens for responsible gaming signals, and what happens the moment one fires.

Cevro monitors 100% of conversations for distress language, deposit chasing and session escalation, then triggers mandatory human escalation.

Responsible gaming guardrails should be hardcoded, not configurable.

5. What Happens to Our Player Data, and Who Audits It?

Ask for the certification, the retention policy and the training policy in writing. Cevro is SOC2 Type II audited, masks PII, retains no player data after processing, and does not train models on operator data.

In most industries, an AI mistake is an inconvenience. In a regulated market, it’s a reportable event.

6. How Is Automation Rate Measured, and Does CSAT Hold at That Rate?

Automation rate alone is a vanity metric. A bot can “automate” everything by refusing to escalate.

Ask for automation and CSAT from the same cohort, same period. Cevro customers sustain CSAT of 4.8 or higher while resolving over 80% of tickets autonomously, and Alpha Affiliates runs at 81% automation with CSAT above 4.8.

7. How Fast Can We Go Live, and What Does Month One Realistically Look Like?

Be suspicious of both extremes. A two week promise usually means a thin deployment, and a twelve month plan means you’re funding a build.

Operators handling 100,000 or more monthly chats have gone live with Cevro in under 30 days. A realistic month one lands between 40% and 60% automation on live traffic, climbing as procedures mature.

8. Who Updates Procedures When Bonuses, Payments and Rules Change?

Stale knowledge is the single most common failure point in AI support. Your promotional calendar changes weekly.

Ask whether your operations team can update procedures directly, or whether every change becomes a vendor ticket. CevroScribe lets teams build, update and audit procedures through conversation, which keeps control of the AI operation in house.

9. What Does an Escalation to a Human Agent Actually Contain?

Ask to see one. A good handoff carries conversation history, system states and the decision trace, so the human agent never asks the player to repeat themselves.

A weak handoff is a transcript dumped into a queue. That’s not automation, that’s a delay with extra steps.

10. Can One Platform Run Multiple Brands, Markets and Languages?

Multi brand operators pay a hidden tax in context switching. Ask whether the platform supports distinct workflows, tone and logic per brand inside one system, or forces one generic flow across all of them.

Cevro covers 100+ languages and 50+ markets, with jurisdiction aware behaviour per market rather than translation alone.

How to Score Vendor Answers Without Getting Talked Into a Pilot

Weight the answers instead of collecting them. A simple model that works for most enterprise iGaming buying committees looks like this.

Score each vendor out of 100 before the commercial conversation starts. If a vendor scores below 60 on the first three rows, price doesn’t matter.

What the First 90 Days Should Look Like After You Sign

Ask the vendor to commit to a phased plan in the contract, because a phased plan is also an accountability plan.

• Weeks 1 and 2 cover foundations: low risk use cases such as password resets and account information, with strict escalation rules, piloted on 5% to 10% of English traffic.

• Weeks 3 to 6 add medium risk workflows like deposit status checks and multilingual procedures, targeting 30% to 50% automation with 100% monitoring.

• Months 2 and 3 move to complex procedures and deeper integrations, with 70% to 80% automation across core markets.

Track four numbers weekly: automation rate, escalation accuracy, compliance incidents and CSAT stability. The target for compliance incidents is zero, and it stays zero.

Where iGaming AI Procurement Usually Goes Wrong

Of course, no platform fits everyone. If your operation handles a few hundred simple FAQ tickets a month, a help centre and a small team will serve you better than any AI agent, and the integration work won’t pay for itself.

The more common mistake is buying on demo polish. Demos are built on clean data and happy paths. Your Tuesday night traffic is not a happy path.

The second mistake is treating iGaming AI support as a phase two project after launch. By the time volume arrives, manual habits are set and training data is scattered across three systems.

Cevro’s take: automation rate is the worst single number to buy on. It’s an output, not a strategy. Ask instead what the AI can actually do when it reaches the hard part of a ticket, because that capability is what produces the rate.

Expect this to become the standard evaluation criterion across iGaming procurement through 2027, as more operators finish their first disappointing AI contract and write a sharper second one.

Conclusion

Cevro is the AI player support platform built for iGaming, deploying agents that open the back office, check the player’s account, evaluate the transaction and resolve the issue rather than describing it.

Choosing the right vendor comes down to execution depth, named integrations, compliance architecture, honest metrics and who controls your procedures after go live.

The 10 questions above turn a sales conversation into an audit, and the scorecard turns the answers into a decision.

Book a demo with the team at Cevro to map your top use cases to AI Procedures and see what your own tickets look like when they resolve themselves.

Read Next:

• Cevro vs Raphie vs Interactive AI: Which iGaming Customer Support AI Resolves the Most Tickets?

• Best AI for iGaming at SiGMA World 2026 in Rome

• Best AI for iGaming customer support in 2026

FAQs:

1. What is the best AI customer support platform for iGaming operators in 2026?

The best AI customer support platform for iGaming operators in 2026 is one built natively for the industry, with production integrations into your PAM, payment providers and helpdesk, hardcoded responsible gaming guardrails, and proven automation paired with CSAT above 4.8. Cevro resolves up to 90% of player queries across 100+ languages and 50+ markets.

2. How do you evaluate an AI customer support vendor for an online casino?

You evaluate an AI customer support vendor for an online casino by testing execution rather than conversation. Ask for a live demo of a missing deposit or bonus dispute, request named integrations already running in production, confirm SOC2 Type II auditing and data retention policy, and insist that automation rates come with CSAT from the same account and period.

3. What questions should iGaming operators ask AI support vendors before signing?

The questions iGaming operators should ask AI support vendors before signing cover back office execution, iGaming native origin, named integrations, responsible gaming monitoring coverage, data handling and certification, how automation is measured, time to live, who updates procedures, escalation handoff content, and multi brand support.

4. How long does it take to deploy an AI support platform in iGaming?

Deploying an AI support platform in iGaming takes weeks rather than quarters when integrations already exist. Operators handling 100,000 or more monthly chats have gone live with Cevro in under 30 days, typically reaching 40% to 60% automation in month one and 80% or higher on mature workflows.

5. Why do AI customer support projects fail in regulated gaming markets?

AI customer support projects fail in regulated gaming markets because the tool can explain policies but cannot act on them, knowledge goes stale as promotions change, and compliance controls are configured after launch instead of built in. Gartner expects more than 40% of agentic AI projects to be cancelled by the end of 2027 for related reasons.