Last week I watched a developer friend spend 3 hours writing proposals on Upwork.
He got one reply. It was asking if he'd work for less.
That's not a talent problem. That's a platform problem.
Here's what actually happens on every freelance platform right now
A client posts a job. Within an hour they have 140 proposals. Most are copy-pasted. A bunch are from agencies running bots. A few are from genuinely talented people who spent real time writing something thoughtful.
The client has no way to tell the difference.
So they pick whoever has the most reviews, or whoever bid lowest, or whoever they saw first. The best fit for the job probably never got read.
The freelancer side is just as broken. You spend hours crafting proposals for jobs you're perfect for and hear nothing. Meanwhile someone with a $5 "I'll write your proposal" gig is spamming 300 listings a day.
This is the system. Everyone hates it. Nobody's fixed it.
What I'm building
Assignly — a freelance platform that assigns jobs instead of auctioning them.
Client posts a task with budget and timeline
Assignly matches the best-fit freelancer based on skills and availability
That freelancer gets notified, reviews the job, accepts or passes
No bidding. No proposals. No mob.
Two things I'm committing to that every other platform won't:
0% commission. No cuts. Ever. A platform skimming 20% off every transaction while adding zero value beyond a listing board doesn't deserve to exist. Monetization comes through optional tools — not taxing your income.
No KYC. No passport scans, no SSN, no selfie with your ID. You're a professional, not a suspect. Trust gets built through work history and reputation on the platform — not by handing over documents nobody asked for.
Where I'm at
The waitlist is live and fully functional — built on Supabase, updates in real time. The core platform is in active development. I'm building openly and will post updates here as it progresses.
I don't have the matching algorithm fully figured out at scale. I don't know exactly how disputes work without a middleman. If you've solved either of those problems before I'd genuinely love to hear how.
What I actually want from you
If you've freelanced anywhere — what's the one thing that made you want to quit the platform?
If you've hired on Upwork or Fiverr — was the problem volume, or just no signal in any of the proposals?
Has anyone tried an assignment-based model before? What broke?
Waitlist is at assignly.edgeone.app. Posting updates here as I build.
Top comments (3)
The matching problem you're describing is real — and it goes deeper than just freelancing. The whole "post a job → get 200 random proposals → pick blindly" model is broken because there's no intelligence in the middle.
Your assignment-based approach is interesting. The hard part will be the matching algorithm — how do you weigh skills vs availability vs past performance vs client preferences? I'd look into how recommendation engines handle cold-start problems (new freelancers with zero history).
I'm working on something tangentially related — instead of matching freelancers to clients, I'm building AI agent teams that handle the work directly. Different angle on the same frustration: the gap between "I need something done" and "it's actually done well."
Two things from your post that resonated: the 0% commission stance is bold and the right call for early traction. And the anti-KYC position will attract exactly the developer audience that hates friction. Smart choices for a v1.
Curious about your dispute resolution without a middleman — that's the hardest unsolved part. Have you considered an escrow-based system with automated milestone verification?
This is exactly the kind of feedback I needed — thank you.
The cold-start problem is the one that keeps me up at night honestly. My current thinking is to lean heavily on skill verification at signup — small portfolio reviews, maybe a short practical test for technical roles — so new freelancers aren’t starting from complete zero. But I haven’t solved it cleanly yet and I won’t pretend I have.
The weighting question is hard too. My instinct is that availability should be weighted higher than most people expect — the best freelancer in the world is useless to a client with a 48 hour deadline if they’re slammed. Skills get you in the pool, availability determines the match.
On escrow — milestone-based is definitely the direction. Automated verification where possible — code passing tests, design files delivered, written content hitting word count — and a human review fallback for anything that can’t be checked programmatically. Not fully built yet but that’s the architecture in my head.
Your AI agent teams angle is genuinely fascinating and I think you’re attacking the same frustration from a completely different vector. The accountability question is what I’d love to understand better from your side — when an agent team delivers something subpar, who owns that conversation with the client? That feels like your version of my dispute resolution problem.
Honestly the problems you’re thinking about — recommendation engines, cold starts, intelligent matching — are exactly what I need people like you around for when I get to that stage. If you’re open to it, would you be interested in being one of the first people I loop in when the matching layer is ready to stress test? Not asking you to commit to anything, just — your brain on this would be genuinely useful and I’d rather have that conversation early than late.
Either way, drop a link to what you’re building. I want to follow it.
Really appreciate the thoughtful response — the accountability question you raised is exactly the hard part. When an agent team delivers something subpar, the audit trail matters: what was the agent instructed to do, what did it actually do, where did it deviate? That's the observability problem I'm solving with TraceHawk — tracehawk.dev
And yes — happy to be a thinking partner when your matching layer is ready. The cold-start + verification problem is genuinely interesting. Following your build.