Parag Jha
heigin.app

HeiGin

We built an HR platform. Two pilots told us they wanted a CRM.

A WhatsApp-native sales CRM for Indian SMB distributors. Six modules planned, one killed after user pilots said the pain was somewhere else.

Built · piloting
heigin.com on a laptop: "Close more deals with less follow-up"
Role
Founding PM and product designer
Dates
2025 to present
Team
Founder team, engineers on React/Next.js with shadcn
Scope
PRD (v1, v2, v3), competitor teardown, RBAC and information architecture, every screen, brand system, landing site
Tools
Figma, Claude, HTML/Tailwind, shadcn/ui
Status
Built and piloting. MVP nearly ready.
Live at
heigin.com
01 · The problem

Attendance and leave weren't the expensive problem. Sales conversations dying in WhatsApp threads were.

HeiGin started as a 6-module HR/finance ERP for Indian SMEs. We designed the whole thing. Then two pilots told us the same thing in different words: attendance and leave weren't the expensive problem. Sales conversations dying in WhatsApp threads were. We stopped a nearly-shipped ERP and rewrote the product as a WhatsApp-native CRM.

The pivot cost us four months of build. It saved us from launching the wrong product.

02 · Constraints and decisions

Five calls that shaped the product.

  1. CRM built as a feature-flagged module.

    Standalone CRM, ERP-only, or combined suite modes, so clients can start where their pain is.

    Rejected:forcing a single bundle and betting on cross-sell.

  2. RBAC with a hard separation of powers.

    Nobody can both approve and execute a payment. Six roles, three hierarchical, three functional.

    Rejected:a simpler Admin/User model that pilots would have failed compliance on.

  3. WhatsApp as the operating surface, not a notification channel.

    Sales reps live in WhatsApp; forcing them into a web dashboard is where every Indian CRM loses adoption.

    Rejected:WhatsApp as an outbound-only integration.

  4. India-deep from day one, not "we'll localize later."

    GST-correct quoting (CGST/SGST vs IGST), HSN/SAC on the product catalog, DND/TRAI compliance, GPS field check-ins, INR-only v1.

    Rejected:a global-first design that ignores what SMB distributors here actually need.

  5. Agent-operated with an earned-autonomy trust ladder.

    The AI reads calls and messages, scores intent, and hands warm leads to the sales team. It doesn't start with full autonomy. Trust is earned per action.

    Rejected:a "full autopilot" pitch that would spook the exact SMBs we're trying to sell to.

03 · The work
RBAC diagram: Global Admin above three separate permission groups (people operations, payroll, finance), with roles implied by org-chart position
The separation-of-powers model. Approve and execute never sit with the same role, and it survived the pivot intact.
System architecture: AI layer on top, six modules in the middle, shared data spine below, immutable foundation at the base
Architecture: modules on one shared data spine, so the CRM could be switched on as a flag.
HeiGin HR manager dashboard overview
The HR build the pilots redirected. Every figure scoped to what the role is allowed to see.
Employee profile with personal details, employment history, and documents
Employee record from the ERP build, each tab gated by the same RBAC model.
Onboarding flow across HR, candidate, and system swim lanes
Onboarding across HR, candidate, and system lanes. One of the pieces that carried over.

Landing site shown separately at heigin.com

04 · Outcome

Where it stands.

  • PRD v3.0 shipped: gap analysis, six-posture actor model, four-level information hierarchy, eight specified flows
  • Impeccable v4 design audit run: identified 1,595 hardcoded hex values, zero ARIA attributes, broken quote builder, non-functional FAB. All logged, most fixed.
  • MVP nearly ready; remaining work is multi-tenant WhatsApp, email and Slack integrations
05 · Reflection

The most expensive design decision I've made is also the one I'm proudest of: killing four months of HR-first work when two pilots pointed elsewhere.

The lesson: pilots are not testimonials. They're supposed to change the roadmap.