For ten years I ran ServeAxis, a California process-serving company that did pretty well by the standards of the industry. We had a steady book of paralegals at twenty-something firms, a network of independent servers across Southern California, an answering service for after-hours, and a creaky web portal that ran on PHP and was older than my marriage.
The thing that finally pushed me to rebuild it from scratch wasn't competition. It wasn't a regulatory change. It wasn't even the realization, somewhere around 2023, that AI was going to do to legal-services workflow what it did to translation, transcription, and customer support.
It was watching paralegals call us seven times for the same status update.
The phone never stopped ringing
The status call is the unit of pain in the legal-services world. A paralegal at a firm files a complaint Monday morning. They order service from us Monday afternoon. Tuesday at 11 a.m. they call for an update. The server hasn't attempted yet. Wednesday at 2 p.m. they call again. The server attempted at 9 a.m., nobody home. Wednesday at 4 p.m. they call again because the attorney just asked. Thursday morning, Thursday afternoon. Friday they call to ask if the affidavit is filed yet.
We had four full-time operations staff. Two of them existed almost entirely to answer the phone and update spreadsheets. The cost per call was almost zero — and there were thousands of them a month.
Every one of those calls was for information we already had. The server had GPS, a phone, a camera. The case had a record. The status was knowable in real time. The information just lived in three places — the server's head, the dispatcher's notebook, the back-office spreadsheet — and the paralegal couldn't see any of them.
The "online portal" wasn't online
We had a portal. Our PHP portal. It showed cases. It did not show real-time status. It did not show attempt photos. It did not show GPS pins. It did not generate affidavits. It did not accept payment. You logged in to see whether the affidavit had been uploaded, which was the same information you could get by checking your email.
Every other vendor in the industry had a portal of roughly the same vintage. The market standard was bad. Nobody could break out because the up-front cost of building anything modern was enormous and the per-case price was low.
What changed was that the up-front cost of building anything modern collapsed. Not zero, but close to it. An AI agent that handles intake is a weekend project now. A real-time dashboard is a week. A GPS-tagged attempt log on a server PWA with offline-first sync is a sprint.
The servers weren't the problem
The other thing that bothered me, the more I looked at it, was the way the industry treated process servers themselves. The dominant model is gig-platform: cheapest server gets the job, race-to-the-bottom pricing, no benefits, no equity, no career path. Server quality is wildly variable because the platforms have no quality enforcement other than star ratings.
We always did the opposite at ServeAxis. We knew every server personally. We onboarded them properly. We trained them. We paid above market. When a server had a bad month we asked what was going on. The cost was higher but the quality was worth it.
If you're going to rebuild the industry, you rebuild it with that model. Servers are partners, not gig workers. Tier 1 is in-house staff with W-2s. Tier 2 is independent contractors with proper agreements, E&O insurance, background checks, and training. Customers buy Servd-branded service with Servd's quality guarantee. They never pick a server — we pick.
Why now
Three things lined up. AI got good enough to handle the intake and dispatch decisions that previously needed a human. The legal industry's tolerance for 1995-vintage software finally cracked — partly because COVID forced firms to actually use the portals they'd been ignoring. And the regulatory environment in California stayed stable enough that a multi-county build is finally lower-risk than it was five years ago.
So we rebuilt. The product is what ServeAxis should have been if we'd had AI ten years ago. The mission is what we've always done — get papers served correctly, return signed affidavits, treat servers as partners — built on infrastructure that doesn't ask the paralegal to make seven phone calls.
That's the story. There is no pivot, no acquisition, no clever financial structure. It is a service-business rebuild with software that is finally as good as the work has always been.