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.
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.
- HRIS, payroll
- ATS, LMS
- Time and attendance
- Scheduling
- Lifecycle events
- Schema normalization
- Behavioral model
- Cross-network signal
- Confidence calibration
- 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:
- Predictive Insights dashboard. A view of at-risk workers, trending signals, recommended interventions, and economic exposure, updated as new data arrives and models recalibrate.
- Engagement tools. The communications feed, recognition flows, pulse surveys, and rewards catalog, used to act on what the intelligence surface recommends.
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
| Component | What it does | Relevant to |
|---|---|---|
| Intelligence layer | Ingests workforce data, runs behavioral models, returns signals | All paths |
| Engagement surface | Communications, recognition, pulse surveys, culture tooling | White-Label, Direct Link / SSO |
| Rewards and economics | Currency model, marketplace, tax categorization | White-Label, Direct Link / SSO |