Image credit: Source: SOFTSWISS press materials. Never imply stock depicts the actual event.
SOFTSWISS sold out its first Tech Race Summit in Warsaw on September 10, 2026, drawing roughly 1,500 iGaming and big-tech professionals to hear a blunt diagnosis: most gambling companies have run an AI experiment, but few have built an AI Ops maturity practice that survives past the pilot. The event, organised by the platform provider behind brands like Bpremium and BGaming, put suppliers from Amazon Web Services, Oracle, Cloudflare, Google, Gcore, Fastly and ScyllaDB on the same stages as operators, payment processors and game studios.
Two numbers from the day frame the gap. Maincard, a payments infrastructure firm, told the summit its AI Ops tooling now cuts incident reaction time to 30 seconds. BGaming, the slot studio SOFTSWISS also owns, said it built a working AI fraud detection system in three months. Both are concrete deployments, not roadmap slides, and both come from companies already inside the SOFTSWISS orbit rather than from outside benchmarking.
The summit ran three parallel tracks (Vision, Solution and Experiment) across more than 40 speakers and 17 partner companies, with representatives from Hub88, The Playa, FinteqHub and TrueLabel alongside the named AI Ops case studies. SOFTSWISS chief technology officer Sergey Kastsukevich framed the event's purpose as breaking down silos: technology knowledge, he said, "should not stay within one company, one sector, or one conference room" but needs to be "shared by technology teams worldwide." A second, fully online edition ran alongside the physical event, giving remote attendees 14 days of access to session recordings.
Why the AI Ops maturity gap persists
An AI pilot is cheap to announce and hard to kill quietly, which is part of why so many gambling operators can point to one. Turning that pilot into AI Ops, meaning AI embedded in monitoring, incident response and fraud detection as a default part of how the platform runs, requires something harder to buy off a vendor price list: a repeatable process that survives staff turnover and does not depend on one engineer's side project.
That distinction matches what platform buyers already learn the hard way when they pick a supplier over building in-house. Operators who chose a third-party platform instead of writing their own player account management system made the same trade the summit's speakers described: buying speed and specialist maturity rather than assembling it themselves, one hire at a time. AI Ops is following the identical curve. The tooling to detect fraud or flag an incident is increasingly available off the shelf. What separates a demo from a production system is whether the operator has the process discipline to act on what the tooling finds.
Security infrastructure has already been through this cycle once. Continent 8 Technologies restructured its own cybersecurity pitch around the same insight this August, converting hosting spend into credits redeemable only for penetration testing, managed detection response and other security services operators kept underfunding. The company's own figures put third-party vulnerabilities behind 60% of sector breaches. AI Ops sits on the identical fault line: the technology exists, but budget and process ownership inside the operator lag behind it.
What the case studies actually show
Maincard's 30-second incident reaction time and BGaming's three-month build cycle are useful precisely because they are specific rather than aspirational. A payments processor that can react to an anomaly in half a minute has moved fraud detection from a human reviewing a dashboard to a system acting on a signal in real time, the same shift already visible in how AI-generated pricing models are starting to set betting markets rather than waiting on a trader.
BGaming's timeline matters too. Three months to ship a working fraud detection system is fast for a regulated product, and it suggests the barrier is not engineering effort so much as organisational will: deciding to prioritise the build, staff it, and put it into production rather than treating it as a research exercise. That same prioritisation question runs through how suppliers distribute content in the first place. Game aggregation platforms already compress months of integration work into API calls, and the AI Ops case studies suggest fraud and monitoring tooling is heading toward the same commoditised, plug-in model.
None of this is universal yet. The summit's own framing, that most companies have AI experiments and few have AI platforms, was delivered by an AWS iGaming specialist speaking to an audience self-selected for interest in high-load infrastructure. The attendees at a sold-out technical summit are not a representative sample of the industry's smaller operators and studios, many of which have neither the headcount nor the cloud spend to run the kind of AI Ops programme Maincard and BGaming described.
The gap is a staffing problem, not a tooling one
The clearest signal from Warsaw is what did not get discussed as a barrier: availability of AI infrastructure. AWS, Oracle, Cloudflare and Google were all present pitching capacity that is, by any measure, abundant and increasingly commoditised for gambling-scale workloads. What separated Maincard and BGaming from the rest of the room was not access to better models or cheaper compute. It was that both companies had already decided AI Ops was a line item worth staffing, funding and shipping rather than a slide in next year's roadmap.
That is a harder problem to solve than buying a subscription, and it is why the maturity gap the summit described is likely to persist past 2026. Suppliers will keep making the tooling cheaper and easier to deploy. The operators who close the gap will be the ones who put a named team and a production deadline behind it, the same way Maincard and BGaming did, rather than the ones waiting for the tooling to get good enough to make the staffing decision for them.