Why We Built Last20

A conversation with Matt Rix, founder of Last20

We recently sat down with ProofStories, a blog that interviews early-stage founders, to talk about the origins of Last20: what sparked the idea, how we validated it, and what we’re learning as we try to help vibe coders actually finish their AI builds.

Below is an edited excerpt from that conversation.

“Most AI tools get you 80% there but it’s the last 20% that kills momentum.”

Where did the idea come from?

It started with my own frustration. I was building with tools like Lovable and Base44, spinning up projects and getting excited. But like a lot of people, I’d get 80% of the way through something and then hit a blocker. A backend integration. A Stripe issue. Some weird bug that made no sense. And that’s where it would stall.

I remember thinking: surely there’s someone who could fix this in 20 minutes? But I didn’t want to deal with Fiverr. Reddit wasn’t consistent. Dev agencies don’t touch small jobs. So I thought, what if there was just a simple task board built for that moment?

Not a new idea. But the positioning felt missing. Lightweight, scoped, and built for AI builders. That’s how Last20 started.

“We assumed the hard part would be devs. In fact loads signed up really quickly.”

How did you validate that there was a real problem?

Honestly? Reddit. I was already spending time in vibe coding threads so I just started replying to people who were stuck. Not trying to sell anything. Just chatting, offering help, and asking questions.

Once we’d built some rapport, I’d say: "If there was a place where you could post this and pay someone £50 to fix it, would you use it?"
And the answer was typically yes.

We thought getting devs would be the hard part. That we'd be flooded with tasks and struggle to get them fulfilled.

We were completely wrong. Devs showed up immediately and now they outnumber vibe coders 3 to 1.

So the challenge flipped. It’s no longer “can we get supply?” It’s: how do we get more builders to post?

“We built the MVP in two weeks. Our first task was completed within three days.”

What did the early build process look like?

I met my cofounder Neil through Reddit. I wasn’t even looking for a cofounder, I just saw his post and messaged him. He came back quickly, we connected, and we moved fast. Within two weeks of that first conversation, we had an MVP live.

We weren’t aiming for perfection. Just enough to test the flow:

• Could someone post a scoped task?
• Could a real dev claim it?
• Could we handle the fix, the handover, the payment?

That happened within three days of launch. We knew it wasn’t polished, but that was a big signal that the core model worked.

“You can post a fix for £1. We’ll pay the dev the full amount.”

How does payment work right now?

Our aim is to be as low friction and as high trust as possible.

You can post for free. You can set your own budget. Payment is held via Stripe and only released when you approve the fix. If you’re not happy, your money goes back. The dev doesn’t get paid unless the work is completed to your standard.

Eventually we’ll add a small platform fee (paid by the task poster), but for now, devs keep 100% of the listed fee.

As an introductory offer, you can set your task budget for £1 and we’ll pay the dev the full fee

“If we get 100 real tasks posted, we’ll know this thing has legs.”

What kind of traction are you seeing now?

We’re averaging 100 - 300 unique visitors a day and signups range from 5 to 30 per day depending on Reddit activity and outreach.

The first organic task was claimed and completed within the first few days of launch. That gave us confidence. But the bigger challenge now is converting users who sign up into people who actually post.

We’re running the platform pretty manually for now, nudging people, scoping tasks, helping facilitate payments. That’s fine. We’re learning a lot.

Our next milestone is simple: get 100 real tasks posted. If we can do that, with quality results and repeat usage, we know we’ve got something worth scaling.

“It’s fixed fee and you only pay when you’re happy.”

What’s been the biggest challenge so far?

Most people we talk to already know what they want fixed, they’re just not sure how to get it done without getting burned.

If you’ve used tools like Lovable, you’ll be familiar with this experience:

You prompt “Fix this button”, it half-fixes it.

You spend more credits.

Then it breaks something else.

You upgrade your plan.

Now you’ve spent £200 and the button still doesn’t work.

You don’t know what it’ll cost. You don’t know when it’s done. You’re stuck in the loop.

Last20 is the opposite of that.

• You say what’s broken

• You describe what “done” looks like

• You set the price

• And you only pay if the work gets done to your standard

And that’s it really, no credits, no subscriptions and no surprises.

It’s fixed fee and it’s the fee you fixed.

It’s not revolutionary. In fact, it’s kind of traditional but it’s the antithesis of the experience that most vibe coders have currently.

“We’re not trying to make noise. We’re just trying to help people finish their apps.”

What’s next for Last20?

In the short term, we’re focused on builder activation: getting more tasks posted and completed successfully.

We’ll start layering in:

• Testimonials and case studies

• Retargeting to people who abandoned mid-post

• Lightweight paid media

• Community-building across Discord, Reddit, and niche dev channels

But we’re not trying to launch with a bang. We’re trying to earn trust by being useful. If we can help people get unstuck, and show that the model works, we’ll scale from there.

“You’ve already done the hard part. Let us help you finish.”

If you’re stuck on your build, whether it’s Lovable, Bubble, Replit, Webflow, or whatever, Last20 is built for that moment.

You don’t need to rewrite your whole stack. You don’t need to hire an agency.
Just describe what’s broken, what “done” looks like, and set a price.

✅ Post a task here and pay only £1 while we’re in beta
💬 Join the Discord if you have any questions or just want to chat