KOL-Craft
AI writing assistant for creator partnerships
# I built an AI tool for creator outreach in 1 week with Claude Code — now I'm stuck on positioning
Hey IH 👋
I just shipped KOL-Craft (https://kol-craft.com) — an AI writing
assistant that drafts personalized first messages for brand-creator
partnerships.
Sharing the build story because the timeline surprised me, and I'd
love feedback before I go full launch.
## The problem
Small DTC brand owners (and the freelancers / agencies who help them)
spend 30+ minutes writing a single outreach message to a creator.
They open YouTube, watch 2-3 videos, scan the channel description,
struggle with the opener, and end up sending something that still
reads generic.
The reply rate on templates is ~2%. On well-personalized emails,
it's 8-12%. But personalization is what eats the time.
That's the gap KOL-Craft fills.
## How it works
1. You paste your brand URL → AI summarizes your positioning + voice
2. You paste a creator's URL (YouTube or TikTok) → AI summarizes
their recent public content
3. AI drafts a personalized first message that references 2+ specific
videos and proposes 3 partnership angles
4. You copy the draft and send it yourself from Gmail/Outlook
Critical design choice: KOL-Craft never sends anything. We're a
writing tool, not a marketing automation platform. This turned out
to matter a lot (see "the hard part" below).
## The build
- Week 0: PRD writing (this took longer than the code, honestly)
- Week 1: Full MVP — 3 codebases (Rust API, Next.js web, Plasmo
Chrome extension), 11 DB tables, AI prompt engineering, Paddle
webhooks
- Week 2-3: Payment processor verification hell (more on this)
Stack:
- Rust + Axum (API)
- Next.js 14 + Tailwind (web)
- Plasmo (Chrome extension)
- Neon (Postgres)
- Anthropic Claude (the AI engine)
- Paddle (payments / MoR)
- Cloudflare Pages + Fly.io (hosting)
- Turborepo monorepo
Solo build. Claude Code did ~80% of the code writing. I did the
PRD, design decisions, prompt iterations, debugging, and product
positioning.
## The hard part wasn't the code. It was payments.
This is the part I want to share most because I haven't seen it
discussed much:
**If your SaaS is anywhere near "outreach", "cold email", or
"marketing automation", payment processors will scrutinize you
before your first user does.**
What happened to me:
- Paddle: initial rejection — "Direct Marketing Services
category not supported"
- Creem.io: pre-approval given, but KYC bounced back without
specifics
- LemonSqueezy: still in queue
What worked for Paddle (eventually approved):
1. Repositioned the entire site from "outreach tool" to "AI writing
assistant"
2. Stripped every mention of "cold email", "outreach automation",
"send emails for you"
3. Anchored my product category against Copy.ai / Jasper / Grammarly
in the appeal email
4. Wrote a strict 14-day no-conditions refund policy
5. Made it crystal clear in 4 places that **users send manually
from their own inbox**
The reposition wasn't dishonest — it was actually more accurate.
But it forced me to articulate the boundary of what KOL-Craft is
(writing tool) vs what it's not (marketing service). That clarity
also helped the product itself.
## Current state
✅ Product live at kol-craft.com
✅ Paddle integrated — first subscription and refund tested end-to-end
⏳ Chrome extension in Google review (1-3 weeks)
❌ Zero paying users yet
❌ AI quality not systematically benchmarked
❌ Cold-start content (templates, blog) not started
## Pricing
- Free: 3 drafts/month
- Starter: $9/mo — 50 drafts, 3 brand profiles
- Pro: $29/mo — 300 drafts, 10 brand profiles, extended partnership briefs
Annual plans get ~22% off.
## What I'm stuck on (would love IH input)
1. Distribution: I have a working product but no audience.
Most of my dev energy went into building. Where would you
start for a B2B tool serving small DTC brands?
2. Positioning depth: Is "AI writing assistant for creator
partnerships" still too narrow? Or not narrow enough? I've
gone back and forth.
3. Pricing: $9 / $29 felt right based on my math (1 sent
reply per month = ROI positive). But I have no signal yet
whether that's the right ladder.
4. Chrome extension dependency: Half my product's value lives
in the Chrome extension that's still in review. Anyone here
shipped extensions before — what's the realistic timeline?
If you do creator outreach, DTC brand work, or affiliate
marketing — I'd genuinely love you to try the free tier and
tell me where the experience breaks.
Open to all feedback. Roast me, please.
---
(Built with Claude Code. PRD-driven. Solo. Happy to share more
specific build details if anyone's curious.)
Comment