Roadmap
What's shipping
before we say it is.
Most HR vendors publish a roadmap that's really a wish-list that never moves. Ours has three lanes and one rule: anything in Now has an owner and an ETA; anything in Next is scoped; anything in Later is on the radar but not committed.
Now
4In flight this quarter (Q2 2026)
-
Workload check-in v1
wellbeingOne-tap weekly mood / load capture, 8-week sparkline, manager-safe aggregate. Replaces the old allocation-driven Saturation heatmap.
-
People-first dashboard tiles
dashboardPersonal status streak, kudos this week, growth score, team Pulse vibe, upcoming Moments — instead of the business-ops KPI grid.
-
Self-host Helm chart v1.0
self-hostProduction-grade Helm chart with external Postgres, Redis, S3-compatible storage, and sealed-secrets integration. Today it's a reference Compose; v1.0 is what we deploy ourselves.
-
Webhook delivery service
ecosystemDedicated delivery queue with HMAC signing, exponential backoff (72h), replay from dashboard, and partial-failure dead-letter for the people events (kudos, log, workload, leave).
Next
4Scoped for Q3 2026
-
MCP server (read-only)
agentsA Model Context Protocol server exposing employees, kudos, status log entries and workload check-ins as resources. Enables external agents (Claude, Cursor) to ground on your tenant data.
-
Real OAuth for Slack / Google / Okta
ecosystemReplace the mock OAuth handshake in Settings → Integrations with the actual flows. Tokens stored in a KMS-wrapped vault.
-
Native iOS + Android apps
mobileNative shells around the PWA with biometric unlock and background sync. Ships via App Store / Play Store; PWA remains the default.
-
Manager digest email
managerWeekly recap delivered to managers: sentiment trend, workload at-risk count, kudos volume, Pulse vibe. Never the raw chat. Opt-in per workspace.
Later
4On the radar for Q4 2026 and beyond
-
Agent-native workflows (write actions via MCP)
agentsOnce the read-only MCP server has settled, extend to write actions — send a kudos, log a day off, mark a workload check-in. Gated behind explicit per-tool permissions.
-
Sentiment summarisation (BYO-LLM)
aiOptional plug-in: route a week of personal status log entries through the employee's own Azure OpenAI or Bedrock tenant for a private retrospective. Data never leaves their network.
-
Marketplace SDK for community surfaces
ecosystemA packaging format (think VS Code extensions) for third-party people-first surfaces distributed on GitHub. Each one declares permissions up front.
-
SCIM for inbound user provisioning
enterpriseNot just outbound identity federation — let enterprises push employee lifecycle events into Pulse from Okta / Entra ID / JumpCloud.
Things we are deliberately not doing.
Every product becomes what it chooses to say no to. Here are the temptations we've said no to, with reasons you can argue with.
- ✗ Time tracking, timesheets, project codes, billing by the hour. Use the tool you already have.
- ✗ Payroll, expense reports, GL/journal exports, tax. Buy payroll software for that.
- ✗ Recruiting kanban, applicant tracking, onboarding workflows, e-signature. Different product, different scope.
- ✗ Performance reviews and OKR cycles. Continuous signal is what we ship instead.
- ✗ Closed-source enterprise edition with extra features. No.
- ✗ LLM-generated chat agents in place of deterministic forms and the local intent parser. Reliability > marketing.
Influence the order.
The lane ordering inside Next and Later is driven by signal — 👍 reactions on GitHub issues, customer meetings, and depth of dependency in the codebase. The loudest way to move something is to open an issue, describe your context, and invite +1s.
Propose something on GitHub Make your work
impossible to miss.
Self-host the open source on your own infra in 90 seconds, or pay us to run it. No demo gate. No enterprise tier with extra features. Export everything in a click — always.