“Vibe coding is the architect sketching a brilliant building. A business rules engine is the building code that makes sure it doesn’t fall down.”
John Grace – North52 CEO
We’re in a moment where everyone with a Copilot / OpenAI / Anthropic subscription and a well-crafted prompt feels like they can build anything. And honestly? They can build a lot. A junior dev with GitHub Copilot can scaffold a Dataverse plugin in an afternoon that would have taken a week two years ago. A business analyst can describe what they want in plain English and watch working code materialize. That’s genuinely impressive, and dismissing it would be foolish.
But there’s a quiet, unglamorous question nobody’s asking at the hackathon: who’s governing the logic after the demo?
A colleague recently told me about a vibe-coded solution that auto-calculated discount tiers for a D365 Sales deployment. It worked beautifully in the demo environment. It also applied a 40% discount to a $2.3 million deal in production because nobody had accounted for a currency field that behaved differently in the Canadian business unit. The developer who built it had moved on to another project. The prompt that generated it was in someone’s personal ChatGPT history. That’s not a horror story designed to scare you off AI. It’s just what happens when logic isn’t governed.
The Cycle We Keep Repeating
Every platform era runs through the same pattern. A new capability lands, people declare existing tools obsolete, and then reality reasserts itself over the next 18 months.
Custom code was supposed to eliminate the need for configuration tools. It didn’t. Power Automate was going to replace everything. It didn’t, and if you’ve ever tried to debug a 47-step cloud flow at 2 AM while an escalation queue fills up, you already know why. Now vibe coding and AI-generated solutions are the new “this changes everything.”
North52 has been in the Dataverse and Dynamics ecosystem for over 15 years. That kind of longevity isn’t inertia. The tools that survive platform shifts tend to be the ones solving a problem the platform itself keeps almost solving but never quite nailing: embedded, governed, auditable business logic. Microsoft ships new automation and logic features regularly. And yet the gap persists, because the problem isn’t about raw capability. It’s about operationalizing logic at scale in a way that non-developers can own and auditors can trace.
Access vs. Embedded: Where Most Enterprises Are Stuck
There’s a distinction that matters enormously and gets glossed over in most AI conversations: the difference between having access to an AI model and having AI built into the workflows where real decisions happen.
Most enterprises are stuck in the left column and calling it digital transformation. Vibe coding often produces left-column solutions dressed up in right-column clothing. It looks like embedded logic, but underneath it’s bespoke, undocumented, and brittle in ways that won’t surface until someone changes a field name or a business rule shifts.
What Vibe Coding Actually Gives You (And Where It Falls Apart)
Let’s be fair about the wins. Vibe coding delivers real speed to prototype. It unlocks flexibility for custom UI work and novel integrations. It lets teams experiment rapidly without waiting in a dev queue. These are legitimate advantages, and they’re why the approach has taken off.
But the gaps are significant, and they tend to show up after the initial excitement fades.
No inherent governance model. Who reviews the AI-generated plugin? Who owns it when the original developer, or the prompt that produced it, is gone? In most organizations the answer is an uncomfortable shrug.
The “works on my org” problem. Vibe-coded solutions frequently end up tightly coupled to one environment’s schema, its specific option set values, its particular relationship names. Moving that logic to another business unit or tenant reveals assumptions nobody documented because nobody knew they were assumptions.
Testing and regression. Business rules change constantly. A pricing model gets updated, a compliance requirement shifts, a new product line launches. Vibe-coded logic often lacks any structured testing framework, which means every change is a manual verification exercise conducted by whoever happens to be available.
Auditability. Regulators and auditors don’t accept “the AI wrote it and it seemed to work” as documentation. In financial services, healthcare, and government, this isn’t a nice-to-have concern. It’s a blocker.
Maintenance debt. There’s a growing and valid concern that vibe coding is creating a new generation of technical debt disguised as innovation. The code that takes ten minutes to generate can take ten hours to understand when something breaks six months later.
What North52 Actually Does (For the Uninitiated)
North52 is a business rules engine that runs natively inside Dataverse and Dynamics 365. It provides a Formula Manager with a low-code formula language for decision logic and Decision Tables that present visual matrices business analysts can read, understand, and own without needing a developer to translate.
Formulas execute in real time on record events, on a schedule, on demand, or cascading across entity relationships. The business logic lives where the data lives, fires when it should, and anyone with the right permissions can read exactly what it does and why. That last part matters more than people think until the first time an auditor asks “how does this calculation work?” and nobody can answer.
Complementary, Not Competing
This isn’t a North52 vs. vibe coding argument. It isn’t North52 vs. Copilot either. It’s about layers in a mature system, each doing what it’s best suited for.
Copilot handles AI-assisted human interactions: suggestions, summaries, draft generation. Vibe-coded custom solutions deliver bespoke capabilities and rapid prototyping for scenarios that don’t fit neatly into existing tooling. North52’s business rules engine serves as the operational backbone, the layer that enforces, calculates, validates, and routes with consistency and traceability.
The organizations getting this right aren’t choosing one approach. They’re using Copilot to help users work faster, vibe-coded solutions to fill genuine capability gaps, and a governed rules engine to make sure the logic that actually runs the business is visible, testable, and owned by someone who will still be around next quarter.
The question worth sitting with isn’t whether AI-generated code can replicate what a business rules engine does. It often can, at least superficially. The better question is whether your organization has a plan for what happens to that logic twelve months from now, when the requirements have shifted, the original builder has moved on, and someone in compliance needs to understand exactly why a customer’s SLA was calculated the way it was.