Contents
8 min read

Smartico MCP Server: Build Segments and Campaigns by Describing Them

Smartico News
iGaming
AI
Written by
Smartico
Published on
October 14, 2026

The Smartico MCP Server lets an AI assistant work inside your Smartico back office. You describe a segment, a campaign, or an automation rule in plain language. The assistant builds it using the same rules the back office enforces, and you review a summary before anything is saved. Developers can also call the same endpoint straight from their own scripts, with no assistant involved.

It's included with every Smartico license. This guide walks through a real task step by step, then covers what the server can touch, how it keeps you in control, what it costs, and the limits to know before your team starts.

MCP in one paragraph

MCP stands for Model Context Protocol, an open standard for connecting AI assistants to outside systems. Anthropic released it in November 2024 and handed it to the Agentic AI Foundation, a fund under the Linux Foundation, in December 2025. Major assistants, including Claude and ChatGPT, support it, and by mid-2026 more than 10,000 MCP servers were reportedly running in production, according to Wikipedia. For Smartico users, the practical upshot is simple: one connection works with whichever assistant your team already uses.

A Monday task, start to finish

Say your retention lead wants to win back casino players who have gone quiet. Done by hand in the back office, that's a segment built across several screens, an audience check, and a campaign flow with two messages. Through MCP, it goes like this.

  1. You type: "Create a segment of players whose total deposits are over €100, whose favorite product is Casino, and who haven't logged in for the last 7 days."
  2. The assistant looks up the matching user properties in your label, picks the operators and values, and combines the conditions.
  3. It runs the segment through dry_run, a validation pass that catches structural errors without saving anything. Then it shows you the logic in plain language.
  4. You ask how many players that is, and the assistant estimates the audience size.
  5. You approve, and the segment is saved.
  6. You follow up: "Draft a campaign that emails these players, then sends a push after 24 hours." The assistant builds the flow, including the steps, the delay, and both messages, and shows it to you before saving.
  7. When you're happy with it, you tell the assistant to start the campaign.

If something looks off at any step, you say so in the chat and the assistant revises it. You still check every condition. The difference is that the clicking, the property hunting, and the operator choices are done for you.

What the assistant can create, change, or only read

Create and update. Segments, automation rules, and campaigns, including their flows, steps, branches, delays, tags, and stop or convert rules. Campaigns can also be cloned, started, stopped, and sped up or slowed down. Beyond those, the assistant can work on messaging templates for push, email, popup, inbox, IVR, and instant messages, plus bonus templates, levels, dynamic reward formulas, and tournament scoring formulas. Label tags, banners and banner placements, avatars, and avatar AI prompts are covered too. So are Forms, which can be sent to players through personal-form campaigns.

Change but not create. Media files. New images, videos, audio, and documents go in through an upload operation, and existing files can be searched, retagged, or deleted.

Read only. User properties, event and activity types, bonus types, deep links, communication gateways, and the gamification catalog: missions, badges, mini-games, raffles, jackpots, clans, and tournaments. The assistant reads these to build the right configuration but doesn't edit them.

Two tools sit outside configuration altogether. One searches Smartico's help articles, academy, FAQ, and public API docs, and answers with cited passages. The other tells you which data-warehouse tables answer a question, with their columns, joins, and example queries, which is useful when a CRM manager and an analyst are working on the same problem.

If you're new to how these pieces fit together, our guides to segmentation and automation rules cover the logic the assistant is working with.

{{cta-banner}}

How it keeps you in control

Handing an AI assistant write access to your CRM is a fair thing to be cautious about. The server is built around that concern.

The assistant acts as a specific person. Every token belongs to a back-office user, and every call respects that user's permissions, so the assistant can only see and change what that person could in the interface. Each call also runs against a single label.

Every create and update can be previewed with dry_run before it commits, and you see a plain-language summary before you confirm. Riskier operations run through a separate tool that isn't marked read-only, which is the signal assistants use to ask for your confirmation before running it.

Access profiles decide which tools each user and token can use, so different people on your team can have different reach. To connect, you either sign in with your Smartico account or use a static token.

That review step matters beyond convenience. Regulators have started asking how AI-driven decisions in iGaming get checked, as we covered in our piece on AI player targeting under regulatory review. A setup where a named user approves each change, within their own permissions, is much easier to explain.

Two ways in: an assistant or your own code

The endpoint an AI assistant connects to is also a standard JSON-RPC 2.0 API over HTTP. Your developers can call it from scripts and backend services with an MCP token, no assistant or special library required. Typical uses are cloning campaigns, bulk-updating templates, and estimating audiences on a schedule.

It isn't a replacement for the Smartico Public API, though. MCP is built for platform configuration and conversational tooling. High-volume player event tracking, real-time gaming transactions, and front-end widget integration still belong to the Public API and the front-end SDKs.

What it costs

There's no extra fee. Access to the MCP endpoint and the configuration tools is part of your Smartico license, and tool calls aren't rate-limited on their own.

Three actions draw on your label's monthly AI usage. Audience estimation and data-warehouse SQL queries are counted by the volume of data they scan. Avatar AI prompt tests count once per generated image. Everything else, including reads, discovery, creates, and updates, uses no AI tokens.

Limits to know before you start

Segments are built on current state. That means lifetime totals such as total deposit count, last-activity dates such as last login, and current attributes such as gamification level. Rolling windows like "deposited 5 times this month" need to be rephrased with the same intent, for example "more than 5 total deposits and last deposit within the last 30 days." Describe your audience that way and the assistant builds it reliably.

The gamification catalog is read-only through MCP. The assistant can use missions, badges, tournaments, and the rest when it builds campaigns and rules, but you still create and edit those in the back office.

It's separate from Smartico's built-in AI Agents. The AI Agents work inside the Smartico back office. The MCP Server lets your own assistant in from outside. Both can search the same product knowledge base.

Coverage is still growing. New tools and object types are added regularly, and an existing connection picks them up automatically, with no reconfiguration. If there's a workflow you'd like exposed, tell your Smartico contact.

Getting started

  1. Check that your back-office user has an MCP access profile with the tools you need.
  2. Connect your assistant by following the MCP Server guide in the Smartico Help Center, signing in with Smartico or using a static token.
  3. Start with something that changes nothing, such as asking the assistant to list your segments and explain one of them.
  4. Then try a real build. "Clone campaign 12345 into a new draft and show me the flow before I start it" is a good first write, because you see the full result before anything goes live.

Want to see it running on a real label first? Book a demo, or if you're at SiGMA World in Rome from November 2 to 5, find us at stand H4-4036. You can book a time at the stand in advance.

Did you find this article helpful? If so, consider sharing it with other industry professionals such as yourself.