Uzski Corp

Corral

Agentic lead acquisition, tailored to what you sell

Corral is an AI agent that finds companies that fit what you sell, emails the person who decides and sorts the replies until a meeting is booked.

Year2026RoleBrand, Product Design & Engineering
Visit Corral ↗
Read the full story

The problem

Lead tools usually split the job. One finds contacts, another sends cold email, and neither knows much about the product being sold. Corral was specced to run the whole thing in one place, starting from what the business actually sells.

What we did

We designed the brand and the product together. You tell Corral what you sell by pasting a site or a deck. The agent searches the open web, reads each company's site and scores the fit. It then finds the person who decides, checks the email and writes a first message that cites the company's own pages. Replies come back to one inbox, the Unibox, sorted and with a draft waiting.

A yes books through Cal.com and opens the deal in an ERP such as Phoenix. The same agent is also an MCP server, so Claude, ChatGPT, Cursor and other clients can ask it which companies fit or who replied. Keys can draft, but only a person can send.

Key decisions

Every score shows its working. A fit score is a short list of weighted checks out of 100. Each check is measured against one line of the product's knowledge base and backed by the page the agent read. Under 80, a company stays out of outreach.

The Open Range identity is sky and dune grass in a run of blues, from #224FD1 to #83A6F3, and it stays outside the work. The product UI has no gradients, glows or coloured panels. The only images inside the app are small range crops on empty states and the sign-in.

Corral
The live landing
Adding a product
The knowledge base
A discovery run
Why it scored
A drafted reply in the Unibox
A meeting booked
The sequence copilot
A live sequence
Integrations
Connecting an MCP client
The sign-in, with a range crop
How it works
How it works on mobile

Have something like this in mind?

Booking for November