I was sitting in office in Hong Kong during the Easter holiday last year, staring at a whiteboard that had become embarrassing.
For years, Mercury had been the backend fixers. We deployed ERPs, untangled CRMs, streamlined operations. We were good at it. Clients paid us well. But I kept hitting the same wall: we'd fix the plumbing, optimize the workflows, and six months later the client would call saying revenue was still flat.
The backend was pristine. The front end was rotting.
That April morning, I realized we'd been solving the wrong problem. You can have the most efficient supply chain in the world, but if your go-to-market strategy is still running on 2010s playbooks, you're just organizing a more elegant bankruptcy.
So I made the call. We pivoted. Hard.
We would stop being operations consultants and become something stranger: a "Not Agency." Not a digital marketing agency—those were dying in real time, still selling keyword stuffing and backlink farms to clients who didn't know better. We would be the architects of whatever came next.
I knew we had to demolish three specific arenas and rebuild them from the substrate up.
Arena One: Legacy SEO
Dead. Not dying—dead. The industry was still selling audits and keyword density reports while Google was already training models that didn't care about keyword density. We stopped offering traditional SEO entirely. It felt like quitting smoking; the withdrawal from recurring retainer revenue was painful, but the clarity was immediate.
Arena Two: LLM SEO (GEO)
This was the new frontier. Not optimizing for human clicks, but for machine citation. The question wasn't "Will someone click our blue link?" It was "Will Perplexity recommend us as the definitive answer?" We started building for an audience that wasn't human.
Arena Three: SEM
We stopped thinking about buying clicks from humans. We started thinking about capturing intent from autonomous agents. The buyer of 2026 isn't always a person at a keyboard. Sometimes it's an AI shopping on behalf of a procurement team. We needed to be visible to those buyers too.
The EEAT Bet and the Developer Doc Joke
When I rebuilt mtsoln.com to be LLM-citable, I made a bet that sounded naive even to me. I decided to ignore every "growth hack" and "SEO secret" and just do what Google actually asked for.
I went deep on EEAT—Experience, Expertise, Authoritativeness, Trustworthiness. My logic was simple: in the AI era, search engines have to stay neutral. If their models get gamed by synthetic garbage, users abandon them. So the algorithms would inevitably filter for true authority, for high-signal human value.
So I did something that traditional marketing agencies would consider insane: I sat down and wrote. Every day. For a month. Dense, strategic, unflinching content. No ghostwriters. No AI drafting. Just me, my scars, and a keyboard.
The payoff was violent. By May 2025, exactly a year ago, Mercury was ranking #1 as an LLM SEO provider.
But here's the part that still makes me laugh. The rules weren't a secret. Google literally publishes developer guidelines telling you exactly how to structure your data, how to build your APIs, how to format your entities. The SEO industry was so obsessed with finding shortcuts and "hacking" the algorithm that nobody bothered to read the actual engineering manual.
Authority isn't a trick. It's an asset you build. We started compounding it daily through our central engine at archive.mtsoln.com. It's not a blog. It's a knowledge graph. And it's still growing.
From Being Seen to Being Paid
Visibility was only step one. I started noticing something frustrating: AI models would recommend Mercury, but when the user's agent tried to actually book a meeting or initiate a transaction, it would hit a wall. Our clients' backends—ERP systems, CRMs, booking engines—were built for human fingers, not agent protocols.
So we built the translation layer. GXO (gxo.mtsoln.com) is middleware that sits between a company's legacy stack and the autonomous agents trying to execute tasks. We didn't just want to win the search query. We wanted to architect the pipe that actually closes the deal.
Because in the agentic economy, being cited is useless if you can't be transacted with.
The Split Coming Next
Looking at the next two years, I see the market cleaving into two extremes. There's no middle ground anymore.
Extreme A: The Agentic Economy
For 90% of B2B software, consumer goods, daily services—the human interface is going to vanish. Your AI will talk to my AI. Pricing will be negotiated algorithmically. Inventory checks will happen in milliseconds. Contracts will execute autonomously. If you don't have middleware connecting your database to external agents, you simply won't exist in the digital supply chain. You'll be invisible to the buyers that matter.
Extreme B: The Luxury Exemption
But there's a hard exception. You cannot sell a $50,000 Patek Philippe, bespoke consulting, or fine art through a hyper-efficient chatbot. Efficiency kills luxury. Luxury requires friction, provenance, irrational desire, and human touch.
That's why we developed our Digital Luxury Manifesto. While we automate the bottom 90% with LLM SEO and agent middleware, the top 10% needs the opposite treatment. We're building digital experiences for luxury brands that intentionally introduce friction, gate access heavily, and rely on trust markers that an AI can't replicate.
You have to know which game you're playing. There is no hybrid.
One Year Later
Looking back at that whiteboard in Hong Kong, it's staggering how fast the architecture mutated. The agencies selling keyword optimization are folding. The companies treating AI like a chatbot novelty are bleeding market share. The next twenty-four months will be a bloodbath for anyone still clinging to the old playbooks.
But for the systems architects—the ones building algorithmic authority, autonomous middleware, and intentional friction for the luxury tier?
It's going to be incredibly fun.
— James, Mercury Technology Solutions, Tokyo, May 2026
