LLC Formation Automation
I built a pipeline that turned "I want a business" into a fully formed LLC — filed with the government, packaged, and mailed to your door. No human touched it after the customer clicked "submit."
// results
$10M/yr
annual revenue
$3M
engineering team managed
100% Python
full stack automation
// the-problem
Starting an LLC is surprisingly painful. You fill out government forms. You wait. You file more forms. You wait more. Then someone mails you a packet and you're officially a business owner. The whole process is manual, slow, and confusing — especially for people who just want to start selling something.
Companies like LegalZoom charge hundreds of dollars to do this for you. But most of the work is just... filling out forms on government websites. Websites that haven't been redesigned since the Clinton administration.
I thought: what if nobody had to touch any of it?
// the-build
The pipeline worked like this:
- 1. Landing page — Customer shows up via Google Ads, enters their info, pays.
- 2. Automated government filing — Python scripts log into state government websites, fill out the EIN and LLC formation paperwork. Every field, every dropdown, every "are you a robot?" checkbox.
- 3. Document generation — Operating agreements, articles of organization, EIN confirmation letters — all templated and generated.
- 4. Physical mail — Connected to a mailing service API that printed, packaged, and mailed the LLC documents to the customer's door.
From "I want a business" to "your LLC package is in the mail" — fully automated. No human in the loop after checkout.
// the-scale
We ranked high on Google Ads for terms like "start an LLC" and "form a business." The funnel was tight: ad → landing page → checkout → automated fulfillment. Customer gets their LLC package in the mail. We generated roughly $10 million a year in revenue at peak.
I also managed the engineering team — distributed across Upwork, about $3 million in cumulative salary spend. Coordinating remote developers across timezones to maintain and scale automation scripts that interact with government websites that break every time someone in the DMV updates a dropdown menu.
If you've ever tried to automate a government website, you know. They don't have APIs. They barely have CSS. And they change their HTML structure on random Tuesdays with no changelog. Keeping that pipeline running was its own full-time job.
// what-i-learned
This project taught me that automation isn't just about writing scripts — it's about building systems that survive contact with the real world. Government websites break. Mail services have outages. Customers enter their name in all caps with a space at the end. You build for the happy path and then spend 80% of your time handling the unhappy ones.
It also taught me to manage remote engineering teams at scale — hiring, coordinating, and shipping across timezones with people I'd never met in person. That skill turned out to be more valuable than any code I wrote.
I've since moved on to building AI agent systems, but the core insight is the same: find the manual process, automate it end-to-end, and get out of the way.