Activity shown in the product preview: Workflow triggered, Task complete, 12 reactions, Credits redeemed.

AI MCP

AI that lives inside your tenant

State a goal in plain English and get a reviewable campaign back in minutes - from AI that runs inside your tenant and never sees the outside world.

Loyalife AI Co-pilot sits inside the tenant boundary, connected to member data, ledger, campaign data, and segments - blocked from the public world outside.
Book a demoSee how it works
From goal to reviewable campaign

Launch campaigns in minutes, not days

Your team states a goal in plain English; the AI plans, drafts the segment, mechanic, and message, and hands back a campaign you can review. Nothing touches your live program until a person approves it. Pick a goal and watch it resolve.

Your tenant · private VPC

Pick a goal

  1. 1

    An operator states the goal

    “Re-activate Gold members who haven't earned in the last 30 days.”

  2. 2

    The model calls scoped tools over MCP - inside the tenant

    suggest_segment(criteria)draft_campaign(segment, goal)forecast_liability(rule)
  3. 3

    Reviewable campaign brief

    • Segment: Gold tier · no earn event in 30 days · live member-count preview
    • Mechanic: bonus-points challenge across POS + bill-pay
    • Comm template drafted for email + WhatsApp
  4. 4

    Routed to Maker-Checker

    A draft, never a change. It is applied to your live program only after a human approves it - and the whole exchange is audit-logged.

Nothing leaves the tenant Nothing applied without human review Every call audit-logged
MCP server · bring your own client

Connect your own AI to Loyalife - inside your walls

Loyalife ships an MCP (Model Context Protocol) server that exposes the loyalty platform as typed, governed tools. Point your own MCP-compatible clients - agents, copilots, or a self-hosted model - at it, all running inside your premises. Because the client, the server, and the data sit on the same network, member data, prompts, and tool execution never cross the boundary. Click a tool to see the data it touches.

Your tenant · private VPC

Your MCP client

agent · copilot · self-hosted LLM

Bring your own MCP-compatible client and point it at the Loyalife MCP server. It plans and calls tools over a local transport - prompts and context never leave your premises.

runs in your perimeter

MCP server

tool-calling surface

Exposes Loyalife capabilities as typed tools. Click one:

Loyalife platform data

Member ledgerin use
Segments & tiers
Rules enginein use
Audit log

forecast_liability(rule) → points, cost

Simulates a rule against real balances to project point issuance and liability before it ships. Pure read - computed in-perimeter.

Prompts + PII stay inside Telemetry hashed Every tool call audit-logged No shared public endpoint
Tenant-isolated · on-prem · audited

Built to clear security review

Encryption, hashed telemetry with keys held on-prem, a private VPN-only cluster, and a hard barrier to anything outside your scope - the controls that turn AI on member data from a fight into a sign-off. Click each safeguard to see what it guarantees.

Your tenant · private VPC

Click a safeguard

No data egress · No cross-customer training · No inference outside your scope

Shared / public model endpoints

Other customers' programs

Out of reach. Your raw member, transaction and reward data is never sent to a third-party model and never trains a shared or another customer's model.

PII encryption

Member PII is encrypted at rest with AES-256 and in transit with TLS 1.3. The data the AI reads is protected end to end - the same record the rest of the platform is built on.

Tenant-isolated, per-program scope Keys held on-prem No shared public endpoint
What the Co-pilot drives

The AI doesn't replace your platform, it operates it

Every lever below is live in your tenant today. The Co-pilot sits on top and drafts, targets, and recommends across the same controls, so a small team ships more without an IT ticket.

Your tenant · private VPC

AI Co-pilot

Drafts, targets, and recommends across every lever below, on your data, inside your tenant.

  • Live

    Visual no-code rule builder

    A drag-and-drop IF / THEN / ELSE engine for earn logic, campaigns, and tiers. Operators build and change rules directly, with no engineering dependency.

    Co-pilot drafts rules

  • Live

    Tier-based personalization

    Multi-tier structures such as Classic, Gold, Platinum, and World, with tier-specific multipliers and benefits, plus automatic upgrade and downgrade as members move.

    Co-pilot tunes tiers

  • Live

    Segment & MCC targeting

    Build segments from demographics, transaction behavior, spend categories, points balance, and MCC patterns, with a real-time member-count preview as you refine.

    Co-pilot builds segments

  • Live

    Missions & gamification

    Multi-task challenges with progress tracking and automatic reward, for example five POS purchases plus one online plus one bill payment earns 1,000 bonus points.

    Co-pilot designs missions

  • Live

    Pay with Rewards

    Proactive auto-redemption against a member’s next qualifying purchase, with configurable validity windows from one to ninety days.

    Co-pilot configures redemption

  • Live

    Referral with abuse caps

    Unique referral codes with automated reward on activation or first transaction, governed by configurable monthly and lifetime abuse caps.

    Co-pilot sets the caps

  • Live

    Multi-channel communications

    Email, SMS, and WhatsApp with event-driven triggers and branded templates, so an earn, a tier change, or a mission completion fires the right message automatically.

    Co-pilot writes the message

Every lever live in your tenant today No IT ticket to ship Co-pilot drafts, a human approves
Next on the AI roadmap · in pilot

The same AI, learning to watch the ledger itself

Beyond assisting your team, the same tenant-isolated AI is being trained to watch the points ledger for abuse and drift, on hashed data inside your perimeter.

Your tenant · private VPC
Designed and in pilot, not yet benchmarked
  • In pilot

    Real-time anomaly detection

    Ten SQL-based detection rules over ledger telemetry, covering fraud patterns, reconciliation drift, expiry anomalies, and phantom transactions.

  • In pilot

    Operates on hashed data

    Detection runs against HMAC-hashed customer data with per-tenant keys held on-prem, so raw identifiers never need to cross your perimeter for risk monitoring to work.

  • Roadmap

    Predictive scoring for high-value programs

    Churn and propensity scoring on the points ledger, positioned for high-value programs and designed to run inside the customer boundary.

Same tenant-isolated AI Runs on hashed data, inside your perimeter Presented as designed, not a benchmarked metric
Zeromember data leaves your tenant
Isolatedper-program tenant scope
Auditedevery AI suggestion logged
On-premdeployment option available
FAQ

What ops and risk teams ask before turning on AI

Live now: the no-code IF/THEN/ELSE rule builder, tier-based personalization, behavioral and MCC segment targeting with a live member-count preview, missions and gamification, Pay with Rewards, referral with abuse caps, and Email/SMS/WhatsApp communications. The AI Co-pilot and its MCP tools are the assistance layer over that same platform - present-tense capabilities, framed by what the tools do rather than benchmarked outcomes. Real-time anomaly detection on the points ledger is in pilot and presented as designed, not yet benchmarked.
Loyalife Co-pilot

We'll have the Co-pilot draft your next campaign on the call

Book a demo