how it works

Three agents. One URL. Ten seconds.

What Scrapitch actually does between you pasting a link and reading your first draft.

01 · the input

One URL is enough.

Scrapitch accepts any public URL: company sites, lab pages, founder portfolios, LinkedIn style profiles.

No CSV uploads. No spreadsheet of leads. No connectors to your CRM.

Paste it once. Pick a use case. Add a sentence about who you are and what you want.

url

https://scrapitch.com

use case

for sales

about you

Operator at Bridgewise, building onboarding for AI products.

your ask

Fifteen minute call about your use case routing.

tone

professional, warm

02 · research analyst

Reads their site like a person would.

The research agent visits the URL and any pages it links to: about, product, pricing, blog, careers.

It extracts structured signal: what they do, who they serve, what they ship, recent launches.

If the site is thin, it pulls what's there honestly. No fabrication.

what_they_do

AI powered personalized cold outreach

who_they_serve

Sales reps, job seekers, grad applicants, founders

recent_signal

Launched v2 with five use cases in May 2026

team_size

1, solo founder

03 · email writer

Three drafts, three frameworks.

The writer uses three proven cold email frameworks: problem, agitate, solution; value first; and curiosity icebreaker.

Each draft has a hard word ceiling: 90, 110, or 100 words. No padding.

Strict no fabrication rule: only claims that came from your input or the scraped site.

problem, agitate, solution87 of 90

The use case routing layer Scrapitch is missing

Multi vertical AI tools usually lose second month users to template confusion. We've seen the pattern twice this quarter.

value first102 of 110

Saw Scrapitch's v2 launch. Sharing what we learned.

Two YC AI companies hit the same multi vertical adoption issues last quarter. Both shipped a routing layer. Happy to share notes.

curiosity icebreaker94 of 100

Quick question about Scrapitch's five use cases

Did you ship the use case picker as part of the v2 launch, or are you experimenting with auto routing based on the URL alone?

04 · scoring judge

Six factors. One score per draft.

The judge grades each draft against a rubric grounded in cold email research.

Personalization counts most. Length, single ask, problem framing, subject line, spam signals fill out the rest.

You see the score and the reasoning. Pick what fits, ignore what doesn't.

Personalization

9.230%

Length

8.520%

Single ask

1015%

Problem framing

8.015%

Subject line

9.010%

Spam signals

9.510%

total

8.97

ready when you are

Three agents. One ten second flow.

Paste a URL. Pick a use case. Send the email.