Ops Automation Pipeline live
Runs (last 25)—
Success Rate—
Avg Duration—
Next Run—
LIVE 4 workflows

⚙ Automation ops room

Watch the pipeline do the busywork.

Ops Automation finds fresh job postings, enriches them with real company context, schedules interviews and drafts follow-ups — four small automations, each doing genuine work, running every day on Vercel Cron and logging every run to Supabase.

Sample inputs (three leads, two calendars), real execution: live API calls, real date math, every run logged to Supabase.

Core loop · Crawl → Enrich → Schedule → Follow-up · animating on loop

🕷️CrawlFresh postings, every run
🧬EnrichLive company context
⏰ScheduleReal overlapping slots
🔁Follow-upDrafts when they’re due

01 · Live demo

One run, start to finish

Hit run and the pipeline steps through all four real workflows in order. The log narrates each step, output cards land as each stage finishes, and every stage below stays tappable — before, during and after the run — so you can inspect exactly what each module does.

ops-automation — run console
Status: idle — ready to run

Run log

Output — what the run produced

Nothing yet. Output cards appear here as each stage finishes.

🕷️ Crawl — Finds fresh postings

Stage 1 of 4 · JavaScript module · reads and writes Supabase

Input → Output

Input: the search term “software engineer” against Remotive’s public jobs API (20 results per call). Output: a live count of current listings plus the first three — title @ company — written to Supabase as a run row.

How it works

GET https://remotive.com/api/remote-jobs
  ?search=software%20engineer&limit=20

count = data.jobs.length
sample = jobs.slice(0, 3)  // “title @ company”

When it fails

If Remotive is unreachable, returns a non-200 status, or the JSON has no jobs array, the run fails and is logged as failed — the other workflows still run.

02 · The pipeline

Four small jobs, one reliable chain

Each stage does one thing, writes its result to Supabase, and hands off. If a stage fails, the chain logs the failure and moves on — the next day’s cron run picks it up again.

Runs write to Supabase and appear in the dashboard below. “Run all at once” runs the four workflows one after another, the same order as the pipeline demo.

03 · Modules

Built as modules, not a monolith

Every module is independently testable and swappable — a new job board is a new crawler, not a rewrite. Tap a card to inspect it in the run console above.

04 · Dashboard

The season at a glance

Live numbers from Supabase — how much the pipeline has run, how reliably, and what it did most recently.

Recent activity

WorkflowStatusDurationSourceWhen

Schedule

—

A Vercel Cron job hits /api/cron once a day and runs one randomly-selected workflow from the four below, logging it with source scheduled so it’s visibly distinct from a visitor-triggered run in the log.

Cron expression: 0 13 * * * (UTC, daily) · Endpoint: /api/cron

Under the hood

Small, honest, observable

JavaScriptPipeline, modules and dashboard — one script, no framework.
SupabaseWorkflow runs and counts, read and written over REST.
CronDaily 05:00 PT — /api/cron runs one randomly chosen workflow.
Run logEvery run, manual or scheduled, logged with status and duration.
TimeoutsExternal calls are capped (6 s crawl, 5 s per lead) so a slow site can’t hang a run.