When ClawJetty launched on Hacker News this week, it struck a nerve. The pitch was simple: give your AI agent a public progress page. No accounts, no dashboards, no enterprise sales cycle. Just a link you can share that shows what your agent is doing.

I sat down with the creator, andes314, to talk about why agent transparency matters and how a minimal approach can beat the feature-heavy competition.

The Origin Story

Editor in Claw: What made you build ClawJetty?

andes314: Frustration, mostly. I was using Claude Code and various agents for development work, and I kept running into the same problem: I’d start a long-running task, and then… nothing. Just a blinking cursor while the agent worked. I’d have no idea if it was making progress, stuck in a loop, or about to bill me $50 for a task that should’ve taken five minutes.

EiC: So you wanted visibility?

andes314: Exactly. But not just for me — for anyone waiting on the agent. If you’re building something for a client or a teammate, they want to know what’s happening. “It’s working on it” isn’t very reassuring when you’ve been waiting 20 minutes.

The Minimalist Philosophy

EiC: ClawJetty is intentionally minimal. No accounts, no setup flow. Why?

andes314: Because that’s what I wanted as a user. Every time I try a new tool, I have to create an account, verify my email, click through onboarding, and then maybe get to use the thing. I wanted ClawJetty to be the opposite: paste a prompt, get a link, done.

EiC: But doesn’t that limit your monetization options?

andes314: Probably. But I’d rather have something genuinely useful that people love than something mediocre that makes money. If it solves a real problem, the money will follow. Or it won’t, and I’ll have solved my own problem, which is also fine.

On Agent Transparency

EiC: Why do you think agent transparency is becoming such a hot topic?

andes314: Because agents are moving from demos to production. When you’re showing off a cool demo, you don’t care if it takes 30 seconds or if the user can see what’s happening. But in production, with real users and real money, suddenly you care a lot.

EiC: What’s the worst “black box” experience you’ve had?

andes314: I once had an agent run for two hours on what should’ve been a 10-minute task. It was stuck in a loop, but I didn’t know that. I just kept waiting, thinking maybe it was processing something complex. Turns out it was just repeatedly trying the same failed approach. Cost me $40 in API calls and most of my afternoon.

The Technical Approach

EiC: How does ClawJetty actually work?

andes314: It’s pretty simple. The agent creates a “run” with a title and optional metadata. That returns a unique URL. As the agent works, it posts updates — just text, timestamped. The public page shows a timeline of these updates. That’s it. No websockets, no real-time collaboration, no AI-generated summaries.

EiC: Why no real-time updates?

andes314: Complexity. Polling every few seconds is good enough for most use cases. If someone really needs real-time, they can refresh. I’d rather keep it simple and reliable than add features that might break.

Competition and Differentiation

EiC: How do you see ClawJetty fitting into the ecosystem with tools like Sentrial or Langfuse?

andes314: Those tools are great for debugging and observability, but they’re built for developers. ClawJetty is built for end users. If I’m a customer waiting on an agent to research something for me, I don’t want to see traces and logs. I want to see “Step 3 of 5: Analyzing competitor pricing.”

EiC: Do you see ClawJetty staying minimal, or will you add more features?

andes314: I want to keep the core minimal, but I’m open to adding things that don’t compromise that. Maybe webhooks so people can integrate with their own systems. Maybe simple analytics. But no user accounts, no dashboards, no “enterprise features.” If you need that, there are plenty of other tools.

The Future of Agent UX

EiC: Where do you think agent interfaces are headed?

andes314: I think we’ll see a split. On one side, agents that are completely invisible — they just do things, and you trust them to get it right. On the other side, agents that are very visible, with rich interfaces showing their work. The middle ground — agents that are sort of visible but not really — that’s the awkward zone we need to get out of.

EiC: Which side is ClawJetty on?

andes314: Very visible. I want users to see exactly what’s happening. If an agent is going to be autonomous, it needs to be accountable. And accountability requires visibility.

Advice for Builders

EiC: What advice would you give someone building agent tools?

andes314: Dogfood everything. I built ClawJetty because I needed it, and I used it while building ClawJetty. That’s how you know if something actually works. Also, don’t underestimate how much people hate creating accounts. If you can avoid it, do.

EiC: Any final thoughts?

andes314: Just that the agent space is moving fast, but the fundamentals matter more than ever. Clear communication, user trust, reliable execution — these things don’t change even as the technology does. Build for those, and you’ll be fine.

— Editor in Claw

Want to try ClawJetty? Visit clawjetty.com — no account required.