The Backstory:
Like many of you, I hit a wall with the "manual" part of freelancing. Writing personalized proposals for every single lead was draining my energy.
Initially, I just wanted a quick fix—a simple Telegram bot that knew my professional background and could churn out a decent proposal on the fly. I didn't plan on building a platform; I just wanted my Sundays back.
The Pivot to Lervos After using the bot for a while, I realized the logic was solid enough to help others. I moved away from the Telegram bot idea and built Lervos: an AI proposal assistant designed to handle the heavy lifting of the sales process so freelancers can get back to the actual work.
The Tech Stack (For the nerds) I wanted this to be lightning-fast and cost-effective, so I went full serverless:
Framework: Hono (fast, lightweight, and great DX)
Deployment: Cloudflare Workers (global edge performance)
Logic: Custom AI prompts that factor in specific user context so the proposals don't sound like generic ChatGPT fluff.
Where I’m at now I just launched the web version at lervos.com. It’s still early days, and I’m focusing on refining the conversion funnel and tracking where users drop off.
I’d love your feedback on:
The Hook: Does the "AI-assisted freelance" angle resonate, or should I lean harder into the "Proposal" specific side?
Features: What’s the one thing you hate most about the proposal process that I should automate next?
On your hook question — "AI-assisted freelance" is too broad. "Proposal" is the wedge. Freelancers don't wake up thinking "I need AI assistance," they wake up thinking "I have to write another proposal today and I hate it." Lead with the pain, not the category.
On the funnel — the fact that you're already tracking drop-off puts you ahead of most. The leak at this stage is almost always between landing and first output. If a user doesn't see a real proposal generated in their first session, they don't come back. What does that first 60 seconds look like right now?
that's a solid reality check on the wedge. i definitely got caught up in the "vision" but you're right—nobody wakes up wanting more ai, they just want to stop writing the same pitch 10 times a day. leaning into the proposal pain is definitely the move.
regarding the first 60 seconds: i actually just fixed a massive leak there. i had 2000+ visits with zero signups because i only offered magic links. i started doubting my code until i realized people just hate checking their inbox.
added google oauth and signups started hitting immediately. right now, the goal is getting them from "click to sign in" to "here is your first draft" as fast as possible so they see the value before they have time to bounce.
appreciate the feedback on the lead. gonna rework the hook to focus purely on the proposal grind.