Principle 01
Specialized tools, not generic APIs
Every agentic action runs through a tool that embodies the business rule it serves. Not a raw endpoint with a prompt taped on.
I've spent decades solving complex business problems with technology for global organizations and their clients. This includes holding a global services role with Microsoft IPTV, structuring profitable multi-year service engagements, and deploying the systems that businesses depend on to generate revenue.
Early in that career, I learned firsthand that most technology investments are operational overhead — a cost center to minimize at best, a business prevention unit at worst. But the systems that generate, track, and collect revenue? Those get funded. Nobody says no to reliable, predictable income.
That insight is why Remergant exists. Everything we build is revenue infrastructure — your sales orders, your invoicing, your payment automation. Not IT overhead. The system that makes you money.
I don't start with requirements documents. I start with two questions: Where's the most friction? and What creates the most value with the least effort?
Often I don't need to ask. It's visible in how people talk about their business, in the workarounds they've built, and in the things they never get done or avoid. My approach has always been inductive — read the business, identify what doesn't work or can be improved, and build the system that resolves what I find.
The result is always more than what was asked for. Not because I add scope, but because the real need is always broader and deeper than the stated request. I define rules, the architecture, and the outputs where none exist — and where no one in the organization could have articulated them. The running automation becomes the specification, and it evolves as the business evolves.
Years of working with executives taught me to speak in their language: expectations and concerns, not deadlines and deliverables.
I learned to plan backward from what matters — anchoring on the dates and outcomes a business cares about, then overlaying the effort required to get there. When the math doesn't work, I say so directly:
"If you want this in place by June, you would have had to start three weeks ago."
That kind of honesty builds trust deeper than any proposal.
Early in my career I built DataMirror's Transformation Server — the Windows NT and SQL Server version of their real-time data replication product, which mirrored transactional data from AS/400 and 3090 mainframes into modern databases. I licensed VBA from Microsoft as the transformation engine — not for scripting, but for its COM+ interfaces and to position the product for the web services architecture that was coming with .NET.
The core mechanism was before-image lookup replication: comparing incoming data against a known prior state to detect what actually changed, then transforming and routing only the deltas. That same pattern drives Remergant's Operational Data Layer today.
DataMirror Transformation Server
Real-time replication from AS/400 and 3090 mainframes into modern databases. Before-image lookup detected change; VBA + COM+ shaped the transformation engine.
Remergant Operational Data Layer
Consolidates business data from every app and cloud service in your revenue cycle — clean, current, and under your control. Same delta-routing pattern, applied to modern cloud services.
At the scale and focus where it actually works.
Everything you see here — the platform, the pricing engine, the ODL, this site — I built with AI. Not in the way the phrase usually lands: not prompts and outputs, not a model left to its own devices.
A working partnership. I bring the vision, the business instinct, and decades of knowing what good looks like; AI is how the work gets built — challenging my assumptions, surfacing blind spots, holding the detail while I hold the direction.
That it works at all comes down to scale and focus. The agentic failures you read about are large, loose, and unsupervised — a model handed generic tools and a long leash. This is the opposite, on purpose. The failure I see most often traces to one decision: giving a language model raw access to tools that simply expose an underlying API. I've never built that way.
The agentic layer gets specialized tools that embody and enforce the business rules — with transactional commits, human-in-the-loop checkpoints where they matter, and a design that produces real, predictable value.
Not autonomous agents running unsupervised — but agents that work the way a trusted team member works, with guardrails, context, and accountability built into every action.
The direction is toward something deeper: systems that don't just run workflows but carry the business logic and operational instincts that took decades to develop. The automation becomes the specification — current, executable, and kept honest by the work itself.
Every agentic action runs through a tool that embodies the business rule it serves. Not a raw endpoint with a prompt taped on.
Transactional commits. Human-in-the-loop checkpoints where it matters. Predictable value, not a gamble on the next token.
Agents that work like a trusted team member — with guardrails, context, and a clear record of what they did and why.
I've always been the person who starts with a blank sheet of paper. Sometimes because an executive asked me to. Sometimes out of impatience. Always out of conviction that I could make it work — and that the act of building is where the real understanding lives.
Remergant is the fullest expression of everything I've learned. The platform, the Operational Data Layer, the pricing model, the way we engage — it's the first time all of it lives in one place, drawing on a career's worth of combining technology with people.
This site is evidence of that. Every page, every system behind it, built in the same collaborative flow state that defines how we deliver for clients.
A conversation, not a sales call. We'll talk about what's broken, what's working, and what would change if it ran itself.
Start a conversation→