Autonomous Upwork job finding
An AI pipeline that captures every Upwork job, scores it against your criteria, writes a personalised proposal, and delivers it to Slack for one-click approval — in under five minutes.
Built on a production-grade stack.
Every job moves through the pipeline independently — one slow client never holds up the rest — and every proposal still goes out under a human's approval.
0min
Poll interval
0
Scoring signals
0%
Human-approved sends
0/7
Never misses a posting
How it works
Fully async and parallelised. You can stop after any one of them and still have a clean, scored job list.
Polls Upwork's GraphQL API every 5 minutes. Deduplication means you never see the same job twice, and never miss a new one.
Every 5 minInstant rule-based rejection: low budgets, bad client history, blocked keywords. No AI cost, no wasted time reviewing garbage.
< 50 msBudget fit, client spend history, skill overlap, competition volume, recency and job type — every job gets a 0–100.
Six signalsReads the job, retrieves your most relevant portfolio pieces via pgvector, and writes a proposal that sounds like you — not a template.
Portfolio-awareA rich Slack message lands with the score, the proposal and the buttons. Copy, edit, regenerate or skip — without opening a new tab.
Slack-nativeWhy it works
Every feature exists because early, specific proposals convert. Everything else is noise.
Early bids win more often. The async pipeline processes dozens of jobs in parallel so you're in Slack before most freelancers open the app.
RAG from your actual portfolio. It reads the client's pain, finds your most relevant work, and writes something specific — not generic.
Only jobs scoring above your threshold reach Slack. Everything below is recorded, visible on demand, and never in your way.
Proposals arrive in Slack, the team works one live board, and decisions sync everywhere instantly.
Run separate profile spaces for different service lines — each with its own filters, thresholds, team and portfolio.
Every job shows its score breakdown, every decision is logged with a reason, and BD performance is tracked against target.
AI proposal engine
It understands the job, retrieves your most relevant portfolio work, and writes something specific enough that the client thinks you wrote it yourself.
# upwork-jobs
Senior FastAPI + PostgreSQL Backend Developer
$5,000–$9,000 · ★ 4.9 · US client · $62k spent
Hi, I noticed you're building a high-throughput ingestion service — exactly the problem I've been solving for three years. At my last engagement I built a similar async FastAPI + PostgreSQL pipeline handling 40k events/min…
Written in 4.1 s · matched 3 portfolio projects · 3.9 min after posting
Isn't this just a template blaster?
No. Every proposal is generated fresh for every job from your actual portfolio — not merged text.
What if I get a bad proposal?
Regenerate it for a different angle in seconds, or edit inline before copying.
Does this violate Upwork's ToS?
You approve every proposal before it leaves your hands. The AI drafts it, you send it.
Pipeline running now
Connect your Upwork account and the pipeline starts immediately. Your first AI-written proposal lands in Slack within minutes.
Sign in