Past Maps: Craig Campbell Bets on Classic Web and Triumphs

In 2022, Craig Campbell, a former engineer at Meta, made a bold decision by selling his e-commerce business aimed at Shopify users. At a time when artificial intelligence was attracting massive investments, his venture capital investors offered him a blank check to start a new project. However, Campbell chose to step away from the flood of investor money flowing into AI to create a website. He founded Past Maps, a service that overlays historical maps onto modern ones. This choice, although risky, has proven to be fruitful.

People generally do not rush into the website sector, especially with the looming threat of Google Zero. However, Past Maps allows users to visualize old maps of a particular area, adjusting the opacity to compare with current maps. The maps come from public sources like the US Geological Survey, but the tools to explore them were developed by Campbell himself. He created these tools to fuel his passion for metal detecting — by identifying modern locations of ancient structures and trails, he could discover new places to search for artifacts.

The site attracts a variety of users, from genealogical researchers to everyday curious individuals, with applications ranging from mapping old oil wells to understanding the evolution of the Duwamish River. Traffic to Past Maps has seen impressive growth, rising from 20,000 to over 300,000 active users per month in three years. The revenue generated is sufficient to support Campbell and his wife, who also participates in the business. Campbell notes that he earns as much as he did when he was an E4 at Facebook, a mid-level engineer.

The primary source of traffic for Past Maps comes from Google search results. Campbell quickly noticed that his site climbed in search rankings when people looked for historical information about places of interest, such as a church frequented by their grandmother or abandoned mine sites in a particular county. By tagging his maps and web pages in a way that Google could understand, he saw a traffic cycle begin to develop.

A web publisher from the old school, 10 or 15 years ago, would likely have relied on display advertising for most of their revenue. You can explore Past Maps with a free account, but to go further, a weekly pass for $9 or an annual subscription for $52 is required. Subscriptions protect Campbell from the whims of fluctuating marketing budgets and an advertising industry largely controlled by Google.

Although AI may devour the open web, Campbell has fully embraced AI tools to manage his business. He explains that he used to spend one to two hours a day handling each service request himself, writing long emails instead of sending a standard response and an FAQ. Now, he uses a local agent model on his desktop to manage first-line triage. His scheduled task runs once an hour — as long as his laptop is on — and has access to his Gmail. This eliminates spam and marketing messages, identifies what needs his attention, and drafts a response. He claims this has reduced his customer service time to about 10 minutes a day.

"I sometimes have unhappy customers," says Campbell. "If they ask me for a refund, it prepares the refund and subscription cancellation request with Stripe. It handles everything, then it notifies me." At that point, he reviews the request, approves or denies it, and checks the message before hitting send.

Campbell also uses AI to help build an OCR — Optical Character Recognition — tool that will work with old maps. "Cartographers are a pain," Campbell jokes. Historical maps pose a particular challenge for existing OCR systems. Labels curve along features like rivers, use inconsistent spacing, and are sometimes crammed together. Campbell found that standard tools failed to analyze these maps. He has had more success with modern LLMs using reasoning, but it’s not as simple as asking an agent to "do the OCR on these maps," he says.

"You still need to bring that human spark into the mix."

Campbell may have turned away from a supposed gold rush in AI, but in doing so, he seems to have created a recipe for a successful online business in the age of AI summaries. When you start with something you are passionate about, create something useful, and share it with others like you, it turns out to be a solid foundation. Campbell's daily life is very different from how one would build and manage a website ten years ago, but the elements that have made his business successful today are deeply human.