10-stage AI pipeline
Language → intent → KB retrieval → risk → compliance → draft → review → action. Every reply is reasoned, not autocompleted.
Built for MGA & UKGC operators
AI player support that passes compliance on the first try.
The conversation-priced AI helpdesk built for iGaming. Telegram, WhatsApp, email, and a web widget — answered by an AI that knows your KB, your jurisdiction, and your responsible-gaming rules.
No credit card required · Live in 10 minutes · Cancel anytime
Compliance-first
Every AI reply screened against your jurisdiction's policy library before it ships.
Realtime everything
Sub-50ms updates to operator dashboards. No polling, no stale state.
10 languages
EN, TR, RU, DE, ES, IT, PT, FR, PL, NL — auto-detected per message.
Built for regulated operators
MGA aligned
Malta Gaming Authority policy library
UKGC aligned
UK Gambling Commission RG ruleset
GDPR ready
PII minimisation by default
SOC2 in progress
Type I expected Q2 2026
What you get
Twelve shipped features. Not roadmap promises.
Every capability below runs in production today. We don't wait for the next quarter to add the one your team needs.
Compliance review queue
Drafts that trip a rule land in a review queue with the reason codes inline. Approve, edit, or block.
Knowledge base with audit trail
PDF, URL, text — embedded with full ingest visibility. Every chunk used in a reply is traceable.
No-code workflows
If a player asks about KYC AND risk is elevated, route to tier-2. Triggers on topic, intent, language, and channel.
Telegram, WhatsApp, email, web
Four channels, one inbox. The widget embeds in any operator site with drag-drop file upload.
10 languages, auto-detected
EN, TR, RU, DE, ES, IT, PT, FR, PL, NL. Player writes in their language; agent reads in yours.
Ticket lifecycle
Conversations escalate to tickets with full SLA tracking. Comments, events, and assignee history audited.
Realtime everywhere
Sub-50ms updates from worker to operator dashboard via cross-process bridge. No polling stagger.
Webhooks + delivery log
Subscribe to 15+ events. Failed deliveries auto-disable after N retries; manual redelivery built in.
Workspace + team RBAC
Workspace admin, agent, viewer roles. Per-brand assignment for multi-license operators.
Analytics across the pipeline
AI confidence distributions, deflection rate, compliance block rate, RG escalation rate.
GDPR posture by default
Sentry user binding ID-only, audit log without PII payload, configurable retention per brand.
How it works
Player to resolution in four steps.
Most replies never reach your agents. The ones that do arrive pre-routed, pre-tagged, and with the AI's reasoning attached.
- 01
Player writes
On Telegram, WhatsApp, email, or your web widget. In any of 10 supported languages.
- 02
AI reasons through 10 stages
Language detect → intent classify → KB lookup → risk score → compliance precheck → draft → judge → action.
- 03
Compliance gate or auto-send
High-confidence safe replies ship automatically. Risky drafts pause for human review with reasoning attached.
- 04
Operator stays in the loop
Realtime inbox surfaces what needs attention. Workflows route, ticket, or tag without manual triage.
One inbox, four channels
Meet players where they are.
Each channel ships with its own auth, attachment, and rate-limit handling. Your agents see one unified timeline regardless of where the conversation started.
Telegram
- Bot per brand, multi-license safe
- Inline media + file attachments
- Dominant in TR / CIS / Eastern EU
- Business API with template support
- Read receipts + typing indicators
- Default channel in LatAm + APAC
- Inbound MX + outbound SMTP per brand
- Threaded conversation matching
- Compliance fallback for regulators
Web widget
- Drop-in script, drag-drop uploads
- Per-brand theming with welcome i18n
- Real-time push from operator dashboard
The AI
An AI that reasons through your rulebook, not around it.
We don't bolt GPT onto your inbox. The orchestrator runs ten discrete reasoning stages — each one inspectable, each one gated by the previous. Your compliance team sees exactly why a reply landed.
- Reasons, not retrieves
- 10-stage orchestrator decides intent, risk, and action before drafting. No naïve nearest-neighbour reply.
- Cites its sources
- Every reply lists the KB articles consulted. Audit trail by default, no extra config.
- Confidence-gated auto-send
- Per-workspace threshold. Below it, the draft becomes an operator suggestion instead of an outgoing reply.
Player support · Auto-reply
en · low risk · KB ✓
Hi, my withdrawal has been pending for 2 days. Can you check what's going on?
Hi! Your withdrawal request from Tuesday is pending an identity verification step. We sent a request for a photo ID to the email on file — once that's approved (typically under 24h), funds will release within your payment method's usual window.
Sources: kb/withdrawal-pending-kyc · kb/payment-window-policy
Compliance
The only AI that compliance officers signed off on.
Generic chat AIs treat regulation as an afterthought. We treat it as the gate every reply has to pass. The rulebook lives alongside your KB, the precheck runs before the draft, and the audit log captures what happened — every time, by default.
When the AI gets it wrong, you have the trace to prove the policy was in place. When it gets it right, you have the record to defend the decision.
Responsible-gaming lexicon
Live in 10 languagesMulti-language phrase library scans every inbound message. Severe signals auto-pause AI auto-send.
Policy precheck stage
Per-jurisdiction configAI drafts run through the policy library BEFORE judge review. Hard-blocks short-circuit the rest of the pipeline.
Per-AI-run audit log
Discriminated by AI vs adminEvery stage decision recorded with input, output, model, and prompt version. 7-year retention configurable.
Operator review queue
Decision capture mandatoryBlocked drafts surface in a dedicated queue with reason codes inline. Reviewer approves, edits, or escalates.
Reach
Ten languages. Auto-detected. No language packs to license.
Player writes in Turkish. AI replies in Turkish. Agent reads in English with a one-click translation. No per-language fees.
ENEnglish
How can I help you today?
TRTürkçe
Size bugün nasıl yardımcı olabilirim?
RUРусский
Чем я могу вам помочь?
DEDeutsch
Wie kann ich Ihnen helfen?
ESEspañol
¿En qué puedo ayudarte?
ITItaliano
Come posso aiutarla?
PTPortuguês
Como posso ajudá-lo?
FRFrançais
Comment puis-je vous aider ?
PLPolski
W czym mogę pomóc?
NLNederlands
Waarmee kan ik u helpen?
Pricing
Pay per conversation. Not per seat.
One player thread within a 7-day window across all channels = one conversation. Agent count never affects your bill.
Starter
Small operators testing the waters
$0/ free forever
- 500 conversations / month
- 1 brand, 3 agents
- All 4 channels live
- 10-language AI
- Community support
Growth
Operators scaling beyond a single market
$0.49/ per conversation
- Volume discounts above 5K / month
- Unlimited brands + agents
- Compliance review queue
- Workflows + webhooks
- Email support, 24h SLA
Enterprise
Tier-1 licensees with bespoke needs
Custom/ annual contract
- Negotiated conversation rate
- On-prem / private cloud option
- Dedicated CSM + onboarding
- Custom compliance rules
- 99.9% uptime SLA
All plans include EU + UK data residency, audit log retention, and GDPR posture by default.
Spin up a compliant AI helpdesk before lunch.
Free for 500 conversations a month, forever. Plug in your KB, hook up Telegram and the widget, ship your first auto-reply in under ten minutes.
No credit card · 10-minute setup · Cancel anytime