← Aman Kumar Limited · RegisterNotes to the accounts · LEAD
Featured · automation · 2026 · Building
Lead-journey automation
An outbound engine that finds and vets prospects and drafts outreach, with a human approving at three gates.
- Role
- Lead-journey workflows, approval gates, results
- Contributor
- Outcome
- An outbound engine that finds and vets prospects and drafts outreach, with a human approving at three gates.
- Links
- Private codebase — described below
- Stack
- Next.js · Supabase · Vercel AI SDK · Trigger.dev · Playwright · Railway
Problem
Outbound was manual and couldn't be measured. The north-star metric: sessions booked per report delivered.
What the system does
A 9-step journey with three human approval gates: targeting → discovery → gate: people → contact finding (only addresses published on a public page, with the source kept) → gate: company → automated company scan → gate: findings → outreach → follow-up.
What I built
- Approve and reject on the pipeline, with an "awaiting approval" filter.
- Outcome logging with undo, and bounce recovery through a backup contact.
- A follow-up due list with nudges.
- A Results tab with the funnel and the north-star metric.
- Do-not-contact enforcement, with unit tests and CI.
Risk factors
- A team codebase: about a quarter of the commits are mine, focused on the approval gates and results.
This is a portfolio, not an offer of securities. No shares exist. Figures are about shipped work.