You Can Vibe Code Anything Now — So Why Do You Still Need North52?

“Without a rules engine, AI outputs are suggestions floating in the ether. With one, they become executed, governed decisions.”

John Grace – North52 CEO

Vibe Coding Is Intoxicating — Until It Isn’t

A consultant I know spent a Saturday afternoon last month spinning up a lead qualification tool for a client using Cursor and ChatGPT. By Sunday evening, it was pulling data from Dataverse, scoring leads, and rendering a slick dashboard. The client loved the demo. The consultant posted about it on LinkedIn. What nobody mentioned was that the scoring logic was baked into a script that only ran when someone clicked a button in the UI, and half the leads entering the system came through an API integration that bypassed the front end entirely. Those leads just sat there, unscored, for two weeks before anyone noticed.

The barrier to building something has genuinely never been lower. That part is real, and it matters.

• Copilot, Cursor, ChatGPT, Replit Agent — these tools let people produce working software at a pace that would have been absurd five years ago.
• For demos, prototypes, and one-off integrations, the productivity gains are enormous.
• But there’s a meaningful difference between giving enterprises access to an AI model and building that model into workflows where real decisions are made. For the Dynamics 365 and Dataverse world specifically, there’s a gap between code that works in isolation and logic that’s embedded where your business actually operates, across every entry point, every record type, every trigger.

A Quick History Lesson: Business Rules Engines Didn’t Appear Out of Nowhere

Tools like North52 exist because of a very specific, very persistent pain point in the Dynamics ecosystem. Business logic needed to live inside the platform, but the people who understood that logic best rarely had the means to put it there.

• Custom plugins were the default answer for years. They worked, but they were brittle, expensive to change, and required a developer for every adjustment — even small ones.
• North52 filled this gap well before “low-code” became an industry category. North52 gave business analysts a formula language. Decision Tables let admins encode complex branching logic without code. Process Genie and Lookup Sets handled scenarios that would otherwise require multiple plugins stitched together.
• The core insight has always been the same: the people closest to the business rules should be the ones encoding them. Not translating requirements into a ticket, waiting for a dev sprint, then testing whether the developer interpreted the logic correctly.
• The vibe-coding movement is rediscovering this same insight, just approaching from the opposite direction. Instead of giving business users a formula language, it’s giving everyone a natural-language interface to code generation. The goal is the same. The durability of the results is not.

What Vibe Coding Actually Gives You (and What It Doesn’t)

It’s worth being specific about where vibe coding delivers genuine value and where it quietly leaves gaps that don’t surface until weeks or months later.

Vibe coding excels at speed to prototype, fast custom UI work, quick API integrations, and demos that land well in a stakeholder meeting. Nobody should dismiss that.

But here’s what tends to get skipped:

Capability	Vibe-Coded Solution	North52
Consistency across entry points	Logic typically fires in one context (UI click, specific flow). Records arriving via API, import, or other channels may bypass it entirely.	Formulas execute at the event level inside Dataverse — regardless of how the record arrives.
Governance & traceability	Who audits a script six months later? Often the answer is nobody, because the person who prompted it into existence has moved on.	Formulas are named, versioned, and discoverable inside the platform.
Deterministic output	AI-generated code may behave consistently, but prompt-driven logic can drift. Regenerating or modifying the prompt can produce subtly different behavior that’s hard to catch.	A Decision Table returns the same answer every time for the same inputs. Full stop.
Runs without user action	Often depends on someone clicking a button or triggering a flow manually.	Fires automatically at the platform event level. Users don’t have to remember to do anything, and they can’t accidentally skip it.
Audit trail	“Who made this decision and why?” is a question most vibe-coded solutions can’t answer cleanly.	The decision, its inputs, and its outcome are documented inside Dataverse.

The crux of it comes down to the difference between having access to a model and having that model’s outputs wired into your operational workflows. Access alone doesn’t produce governed outcomes.

The Controversy: Are Business Rules Engines Legacy Tech?

Some voices in the community argue that tools like North52 are a relic. Power Automate plus Copilot plus custom code covers everything now, the argument goes. Why maintain another layer?

It’s a reasonable question, and it deserves a direct answer rather than hand-waving.

• Many organizations gave employees Copilot access and waited for organic adoption. What actually happened in a lot of cases was inconsistent usage, ungoverned outputs, and solutions disconnected from the processes they were supposed to support. Access without structure didn’t produce the transformation anyone expected.
• Vibe-coded solutions have a particular failure mode: they create what you might call shadow IT 2.0. Things work fine until the person who built them leaves the organization or simply forgets how the prompts were structured. There’s no institutional memory in a one-off script.
• A business rules engine isn’t a feature you bolt on. It’s infrastructure. It’s the decision layer of the platform — the place where “if this, then that” logic is centralized, visible, and maintainable by the people responsible for it.

The real question isn’t “North52 vs. AI.” It’s whether business-critical logic should live in discoverable, auditable, maintainable rules or in scattered code and prompts that nobody owns after the first quarter.

The Future: North52 and AI Are Complementary, Not Competing

The most useful framing here isn’t either/or. It’s a pipeline.

Think of it as a sequence where each stage does what it’s best at:

1. AI generates a recommendation. A model scores a lead, flags a risk, classifies a case, or suggests a next action.
2. North52 acts on it. A Decision Table takes that AI-derived field as an input and produces a deterministic output: route this case to the right team, escalate this opportunity, apply this discount tier, trigger this notification.
3. The decision is documented inside Dataverse. What happened, why, and based on what inputs — all traceable.

Without that middle layer, AI outputs are recommendations that may or may not get acted on. Someone has to see them, interpret them, decide what to do. With a rules engine in the loop, those recommendations become executed, governed decisions that run consistently whether it’s 3 PM on a Tuesday or 2 AM on a holiday weekend.

Decision Tables are particularly well-suited to this because they’re already built to handle multi-variable branching logic. Adding an AI-scored field as one more input column doesn’t require rearchitecting anything. It just extends what’s already there.

The organizations that will get the most out of AI in Dynamics 365 probably won’t be the ones with the most Copilot licenses or the most vibe-coded prototypes. They’ll be the ones that figured out where AI-generated insight ends and operationalized decision-making begins — and built the connective tissue between the two.