Why I built leadrack

leadrack started as a tool for myself. My other product, Workalong, sells job ad pages, so our best prospects are companies that posted a job in the last few days. That signal goes stale fast, which meant every week I sat down and ran the same search by hand: scan the job boards, pick out the companies that actually fit, dig up a contact, write a first email. It worked, but it was hours of repetitive judgment calls that an agent with a clear brief could make just as well. I went looking for a tool that did exactly that and found either bloated sales platforms, tools that work with (stale) dataset or scrapers that dump ten thousand (mostly irrelevant) rows into a spreadsheet.

So I built it. leadrack runs Workalong's prospecting today, which is also why there's a quote from us on the landing page. Every day the board gets fresh companies that just posted a job, with the ad attached as evidence and a first draft already written.

It's not a CRM

The first question everyone asks, so: no, leadrack is not a CRM and isn't trying to become one. It sits in front of whatever you already use. Agents hunt across sources, a judge scores every candidate against your brief, and only leads that clear your bar land on the board, each with evidence attached. The board is where leads get qualified and get their first touch. Once something turns into a real conversation, you move it into your CRM. Everything on the board is reachable over the API, so that handoff is a small script rather than a migration project.

Your agents can run the whole board

When most tools say "integration" they mean you can push rows in. In leadrack, everything a teammate can do to a lead, an agent can do over the API (or MCP, if your agent speaks that instead). Adding leads, enriching contacts, moving cards through stages, reading timelines, drafting and sending email. The UI and the API call the same service layer, so there is no gap between what a human can do and what your agents can.

You choose the depth. Point a Clay enricher at the board, let a Claude agent triage the new column every day, or hand over the whole pipeline and just read the timeline. Every change is attributed to whoever made it, human or agent, so you can always tell who did what.

Thirteen sources at launch

Agents can currently hunt across thirteen sources: job postings, Reddit, Ask HN, launch sites, funding announcements, SEC Form D filings, GitHub, podcasts, Bluesky, GDELT's news firehose, any RSS feed, plus a general and a deep web scout. You describe your ideal customer once and each agent translates that brief into its own corner of the internet. More sources are on the way. If your leads live somewhere we don't cover yet, tell me, there's a good chance it moves up the list.

Automate as much as you want

Including the sending. Connect your own SMTP mailbox (Fastmail works great, anything that speaks SMTP does) and email goes out from your actual address. New agents start in draft mode, where every accepted lead gets an unsent draft waiting in your inbox. Once an agent has proven itself through a handful of approved sends, you can flip it to send mode, and leads that clear a second, higher score threshold get emailed without anyone watching. The bar for your outbox is deliberately higher than the bar for the board.

Fully set up, you write a brief and the agents find, qualify, and contact leads automatically. That is how it runs for Workalong.

Just getting started

The bones are in place: the board, the agent fleet, the judge, the sending guardrails. It does real work for a real company every night, and the list of things I want it to do is still much longer than what it does today. If this sounds like the tool you've been looking for, try it and tell me what's missing. Right now every piece of feedback moves the roadmap.