Platform feature flags

How InsightHire platform admins enable capabilities globally and per tenant, with the complete slug catalog.

InsightHire uses a three-layer gating model for premium capabilities. Customer org admins see features only when all relevant layers pass.

Gating layers

flowchart TD
  A[Feature request] --> B{Layer 1: Platform kill switch ON?}
  B -->|No| X[Hidden everywhere — PLATFORM_OFF]
  B -->|Yes| C{Layer 2: Org grant?}
  C -->|No| Y[Hidden for org — NOT_GRANTED]
  C -->|Yes| D{Layer 3: Org admin toggle?}
  D -->|No| Z[Grant OK but org opted out]
  D -->|Yes| OK[Feature active — GRANTED]

Layer	Storage	Who controls	Example
1. Platform kill switch	`platform_integration_settings.enabled`	InsightHire platform admin	Global off for beta feature
2. Per-tenant grant	`org_feature_grants.granted`	Platform admin per org	Pilot customer gets `custom_pipelines`
3. Org admin toggle	Org-scoped settings tables	Customer org admin	Culture profile `enabled` flag

Customer-facing gates use isFeatureGrantedForOrg(slug, organizationId) (layers 1 + 2). Layer 3 is feature-specific — e.g. culture fit requires grant and org profile enabled.

Missing grant row = NOT granted (conservative default). Revokes set granted: false instead of deleting rows for audit trail.

Platform admin UI

InsightHire Admin → Integrations & Features (insighthire-admin/app/integrations/page.tsx):

Global toggle — trpc.platformAdmin.toggleIntegration({ slug, enabled }) flips layer 1
Tenant Access tab — appears only for feature-flag slugs; lists all customer orgs with grant toggles
Grant toggle — trpc.platformAdmin.toggleOrgFeatureGrant({ organizationId, slug, granted })

Supporting procedures:

Procedure	Purpose
`listIntegrationSettings`	All slugs with `grantedTenantCount`
`listOrgFeatureGrants`	Org list for one slug (search by name/domain)
`toggleOrgFeatureGrant`	Upsert grant + Redis cache invalidation
`seedIntegrationDefaults`	Bootstrap `platform_integration_settings` rows

Cache invalidation publishes on Redis channel feature-flags:invalidate so all API replicas drop in-process cache within milliseconds (30s TTL fallback).

FEATURE_FLAG_SLUGS (code constants)

From insighthire-api/src/lib/feature-flags.ts:

Constant	Slug	Category
`MEETINGS_MICROSOFT_TEAMS`	`meetings_microsoft_teams`	Meetings
`MEETINGS_GOOGLE_MEET`	`meetings_google_meet`	Meetings
`CULTURE_FIT_SCORING`	`culture_fit_scoring`	Hiring intelligence
`INSIGHTHIRE_LIVE_INTERVIEWS`	`insighthire_live_interviews`	Meetings
`TALENT_NETWORK`	`talent_network`	Hiring intelligence
`CUSTOM_FIELDS`	`custom_fields`	Hiring intelligence
`JOB_GOVERNANCE`	`job_governance`	Hiring intelligence
`CUSTOM_PIPELINES`	`custom_pipelines`	Hiring intelligence
`STRUCTURED_INTERVIEWS`	`structured_interviews`	Hiring intelligence
`CUSTOM_REPORTS`	`custom_reports`	Hiring intelligence
`DEI_COMPLIANCE`	`dei_compliance`	Hiring intelligence
`CANDIDATE_SURVEYS`	`candidate_surveys`	Hiring intelligence

Full slug catalog (platform_integration_settings)

Seeded by platformAdmin.seedIntegrationDefaults in platform-admin.ts:

ATS & HRIS

Slug	Name	Default enabled
`greenhouse`	Greenhouse	Yes
`lever`	Lever	No (coming soon)
`workday`	Workday	Yes
`bamboo`	BambooHR	No (coming soon)

Job boards

Slug	Name	Default enabled	Notes
`indeed`	Indeed	No	Per-position email parsing today
`linkedin`	LinkedIn	Yes	OAuth + Apply Connect

Meetings & calendar

Slug	Name	Grant-gated
`meetings_microsoft_teams`	Microsoft Teams scheduling	Yes
`meetings_google_meet`	Google Meet scheduling	Yes

Background checks

Slug	Name	Default enabled
`accurate`	Accurate Background	Yes

Per-org credentials in Settings → Integrations; not the same grant model as hiring intelligence flags.

Hiring intelligence (per-tenant grant tab)

Slug	Name	Customer settings surface
`culture_fit_scoring`	Culture Fit Scoring	Settings → Culture
`talent_network`	Talent Network	Network nav, reject release
`custom_fields`	Custom Fields	Settings → Custom Fields
`job_governance`	Jobs Governance	Approvals, Intake Forms, Requisitions
`custom_pipelines`	Custom Pipelines	Settings → Pipelines, kanban, tasks
`structured_interviews`	Structured Interviews	Settings → Interview Kits
`custom_reports`	Custom Reports	Report builder, dashboards
`dei_compliance`	DEI & Compliance	Settings → Compliance
`candidate_surveys`	Candidate Surveys	NPS on stage change, analytics

Additional grant slug (admin UI set)

Slug	Name	Notes
`insighthire_live_interviews`	InsightHire Live	Native Daily.co live interviews

UI gate reason messages

getFeatureGateForOrg returns:

Reason	Meaning for support
`PLATFORM_OFF`	Feature not launched globally — enable kill switch first
`NOT_GRANTED`	Global on, but org not on pilot list — grant tenant access
`GRANTED`	Layers 1–2 pass — check layer 3 org toggle if still hidden

Settings tabs hide silently (no disabled stub) when grant missing.

Typical enablement workflow

Platform admin enables global switch (toggleIntegration)
Platform admin grants pilot tenants (toggleOrgFeatureGrant)
Customer org admin configures feature in Settings
Recruiters see new nav/tabs within ~30s (cache TTL worst case)

For soft launches (talent_network, culture_fit_scoring), keep global OFF until ready, then grant select tenants before global enable.

Slugs vs integrations

Not every platform_integration_settings row uses per-tenant grants:

Model	Slugs	Gate
Feature grant	`custom_fields`, `job_governance`, …	`org_feature_grants`
Per-org connection	`greenhouse`, `linkedin`, `accurate`	Org integration credentials in Settings
Env + platform config	`meetings_*`	Grant + recruiter OAuth

The admin UI Tenant Access tab renders only for FEATURE_FLAG_SLUGS set members.

Customer-facing documentation map

Slug	Doc
`custom_fields`	Custom fields
`job_governance`	Job governance
`custom_pipelines`	Pipelines & automations
`structured_interviews`	Interview kits
`culture_fit_scoring`	Settings → Culture (see Roles & permissions)
`dei_compliance` / `candidate_surveys`	Settings → Compliance

Requesting flags (customers)

Contact InsightHire support or your CSM with:

Organization name and domain
Desired slug(s)
Target go-live date

Platform team enables layer 1 (if needed) and layer 2 for your tenant.

Settings overview — which tabs are gated
Roles & permissions — flags vs org roles

Platform feature flags

On this page