Back
docs
Getting started

Infrastructure overview.

MustardHub is behavioral workforce intelligence infrastructure. It ingests transactional, historical, and behavioral data from workforce systems, applies predictive behavioral models, and surfaces forward-looking insights and engagement tools. For HCM, compliance, and workforce platforms, that means adding a predictive intelligence layer to their product surface and delivering actionable insights to their customers. For those platforms’ customers, it means tools to see around corners, retain workers, improve culture, and reduce turnover costs.

The core problem

Transactional workforce systems (HRIS, payroll, ATS, LMS, time and attendance) record outcomes after they happen. An employee leaves; the system records the separation. A retention problem appears; the system has no signal until it’s already a fact.

MustardHub sits upstream of those outcomes. It reads behavioral patterns in historical and ongoing transactional data, models what those patterns predict about future behavior, and returns actionable intelligence before the outcome materializes. The infrastructure’s structural advantage is cross-network signal: behavioral patterns aggregated across the full MustardHub partner network inform predictions for any individual company, producing accuracy no single customer’s or partner’s dataset can generate alone.

Each partner’s data remains strictly isolated within their credential scope. Cross-network signal is derived from aggregated, anonymized behavioral patterns across the network, not from any exchange or commingling of raw data between partners or customers.

Two product layers

MustardHub ships as two distinct but complementary products. Partners may deploy one or both depending on their chosen integration path.

Layer 01
Intelligence layer
Ingests workforce data, runs behavioral models, returns predictive signals. Available to all partners regardless of integration path. The core of the Embedded API path.
Layer 02
Engagement surface
Communications, recognition, rewards, pulse surveys, and culture tooling. Deployed as the customer-facing product in White-Label and Direct Link / SSO paths.

Intelligence layer

The intelligence layer is the behavioral modeling engine. It accepts workforce data across multiple categories, normalizes it to MustardHub’s schema, applies cross-network behavioral models, and returns calibrated predictive signals.

Data inputs
Workforce records
  • HRIS, payroll
  • ATS, LMS
  • Time and attendance
  • Scheduling
  • Lifecycle events
→
MustardHub
Behavioral intelligence
  • Schema normalization
  • Behavioral model
  • Cross-network signal
  • Confidence calibration
→
Outputs
Predictive signals
  • Risk index
  • Churn trajectory
  • Economic exposure
  • Recommended actions

Signals are grouped into four categories (risk, economics, operational, and action) and are addressable independently so partners can surface only the signals their product is ready to act on. See Embedded API for the full signal catalog and ingestion model.

Engagement surface

The engagement surface is the customer-facing product deployed in White-Label and Direct Link / SSO paths. It includes the tools operators use to act on the intelligence the platform surfaces: communications, values-based recognition, peer-to-peer and company-to-employee currency transfers, pulse surveys, and culture management tooling.

The engagement product runs inside the same account architecture as the intelligence layer: the same workspace, the same member roster, the same hub and organizational structure. Intelligence and engagement are coupled in the product; a partner deploying White-Label or Direct Link / SSO gets both.

See White-Label for the full engagement feature inventory.

How customers use it

A customer on MustardHub is typically an HR, operations, or people-function business leader. They interact with MustardHub through one of two surfaces:

Workspace model

Every customer lives inside a Workspace, MustardHub’s tenant container. A Workspace holds the member roster, hub structure, values configuration, funding balance, and all engagement history for a single organization. Partners interact with Workspaces through the API or through the admin controls on their chosen path.

Workspaces are isolated per partner credential. A partner credential cannot access another partner’s Workspaces. Within a partner’s scope, multi-tenant deployments use the Workspace’s tenant_id to isolate data and signals.

Platform components at a glance

ComponentWhat it doesRelevant to
Intelligence layerIngests workforce data, runs behavioral models, returns signalsAll paths
Engagement surfaceCommunications, recognition, pulse surveys, culture toolingWhite-Label, Direct Link / SSO
Rewards and economicsCurrency model, marketplace, tax categorizationWhite-Label, Direct Link / SSO