AI & Brand · MCP & API
Your retention stack as governed tools AI agents can call.
Loyalty, campaigns, members, and segments — made available as MCP tools that AI assistants like Claude, Cursor, and Copilot can discover and use under your governance. MCP is the open standard that lets AI assistants securely use your tools — so this is no fragile custom integration, and no access keys pasted into a chat window.
The problem
Your team is adopting AI assistants faster than your systems can safely open up to them.
Fragile one-off integrations
Every new integration means custom-built connections that someone has to keep fixing as things change.
Ungoverned access
Handing an AI assistant a raw access key usually means handing it everything — no permission limits, no data boundaries, no record of what it did.
Assistants can't discover what you do
A traditional integration is invisible to an AI assistant until someone hand-codes it in. Your capabilities stay locked away.
How it works
14 inbound tools, every call governed.
Flash speaks the open MCP standard at a single secure address (/api/mcp). An AI assistant first asks what Flash can do (tools/list), then calls the specific tool it needs (tools/call) to act — and every call is authenticated, permission-checked, kept to your own data, and logged.
Inbound MCP tools
14 totalEach tool has a clearly defined set of inputs and a strict boundary that keeps it to your own data. The same capabilities are also available as a standard API for systems that aren't AI assistants.
Authentication
Authorization: Bearer fl_live_••••••••Keys are encrypted at rest and shown in full only once, when you create them. Each key carries its own permission scopes and can be revoked at any time. Full access to every tool (/api/mcp) requires a key with all permissions.
Headless REST API
Run Flash end-to-end from your own code: manage members, orders, the product catalog, coupons and tags, plus audience segments, campaigns and automations — build a campaign draft by API, then a human reviews and sends. Load large files with async CSV/JSONL import (members and orders, with a job you poll — no small batch cap), and validate against real behavior first with test-mode keys whose data stays out of your live reports and can be purged in one call.
Every key is governed: per-key rate limits with standard headers, one-click secret rotation, and a usage panel showing volume, error rate and your busiest endpoints.
One governed path, every agent
Whether the caller is Claude, Cursor, or Copilot, every tool call passes the same four checks before it touches your data.
Identity
secure API key
Permission
8 fine-grained scopes
Isolation
your data, kept separate
Audit
every call logged
Permission scopes
Two-way MCP
The same standard, the other direction.
In one direction, AI assistants call Flash. In the other, Flash connects out to your other customer-data tools — pulling each member's details to personalize their email as it's built. The whole connection is built in-house, with no extra third-party software to depend on.
Automatic fail-safe
backs off when a tool is down, retries as it recovers
Controlled load
never overwhelms your connected tools
Healthy connections
reuses healthy connections, routes around bad ones
Always-valid access
refreshes credentials automatically before they expire
Into email personalization
Nine ready-made connectors (including AWS Personalize, Recombee, Algolia, Bloomreach, and Shopify) feed personalized content into each email as it's built. If one source is unavailable, that recipient simply falls back gracefully — the email still goes out, never broken.
This connection is built and ready, but switched on only once you add your credentials — until then it simply stays disconnected and emails fall back gracefully. We don't present it as proven at large scale yet.
Why it's different
Beyond traditional integrations and Zapier.
A traditional integration is a fixed, custom-built connection. Zapier runs pre-set automations when something happens. MCP is the open standard AI assistants discover and use on demand— and it's what the industry is consolidating around.
Typical approach
Traditional integrations
Hand-built connections an AI assistant can't discover; you maintain them.
Flash, by design
Self-describing MCP tools that AI assistants find and use on demand.
Typical approach
Zapier & similar tools
Pre-set triggers, not on-demand action by an AI assistant.
Flash, by design
AI assistants use your capabilities in the moment, under your permissions and audit log.
Typical approach
Raw access keys in a chat
All-or-nothing access, no boundaries, no record.
Flash, by design
Permission-scoped keys, your data kept isolated, and a logged audit trail on every call.
AI & innovation
On the standard the whole industry is converging on.
MCP — introduced by Anthropic, donated to the Linux Foundation's Agentic AI Foundation, and now generally available inside Microsoft Copilot Studio — is becoming the standard way businesses make their capabilities available to AI assistants, with governance and audit built in. Flash was built around MCP from the start, in both directions.
AI-discoverable capabilities
AI assistants discover what Flash can do on their own, instead of waiting for someone to hand-code a connection.
Governance built in
Secure identity, fine-grained permissions, data isolation, and a full audit trail — exactly the enterprise gap 2025's MCP work closed.
Built in-house, fewer parts to trust
The fail-safes, load controls, and credential handling are all built in-house — fewer outside parts to depend on.
What changes for the business
Your retention capabilities become safe, ready-to-use tools for the AI assistants your team already uses.
14 tools
loyalty, members, coupons & more, AI-ready
8 scopes
fine-grained permissions, enforced per key
Two-way
AI assistants call in, Flash connects out to your data tools
Your retention stack as governed tools an AI agent can call.
Related reading
Keep exploring the pages most related to this one.
Intelligent Decision
A next-best-action: who, which offer, which channel, when.
Read more CapabilityConsumer World Model
The core technology under every decision: a Behavior Model that learns how your members actually behave — trained per brand on their own event stream, honesty-gated against a transparent baseline — plus a Digital Twin that simulates what they'd do next, rehearsing campaigns (and the option of doing nothing) before a message goes out. Together, one complete, living model of your member base.
Read more CapabilityDashboard
Real-time scan-to-redemption data as KPI trends, funnel diagnostics, and a rule + LLM next-best-action panel — plus retention curves by signup cohort and a 14-day revenue outlook that's honest about being a projection, not a commitment.
Read more CapabilitySoTag for Slack
The Slack agent that hands governed, audited numbers to Claude in-thread: daily anomaly scan + human-approved coupon-send. Early access.
Read more CapabilitySoClaw — Hire your AI employees
Hire AI employees for your loyalty program — autonomous agents that each hold a real job: a Loyalty Manager that wins back at-risk customers, an Order Manager that clears the receipt and order review queue, a Risk Manager that protects points from fraud, and a Marketing Manager that turns a goal into a ready campaign. Each works one case at a time within limits you set and logs every decision; you manage them like a team with goals, budgets, tasks and feedback. Off by default, capped, instantly pausable. Early access.
Read more CapabilityAI Governance & Approvals
Every AI action runs on a per-action authorization gear — automatic, human review, or forbidden — with an approvals console, a complete decision log, one-click undo, and automatic demotion to human review when an action starts misbehaving.
Read more