I recently got my hands on some very fresh bits of Business Central functionality, never seen by the public before. This is the AI-driven future we have all been promised. Check out the video if you want to see the future:
In this video, Erik gets his hands on what appears to be a futuristic build of Business Central — one that dynamically generates entire application experiences on the fly using AI. He walks through several jaw-dropping demos where he simply types what he wants into the Tell Me search bar, and Business Central responds with fully-formed pages, data, and even business logic for domains that have never been built as extensions. The implications are enormous, and Erik is equal parts excited and unsettled by what he’s seeing.
A Business Central That Just… Knows What You Want
On the surface, the demo starts normally enough — Erik is on the Customer List page in what looks like a standard Business Central environment. But the moment he opens the Tell Me search bar and starts typing, it becomes clear this is anything but standard.
Erik types “animal hospital management” and is immediately presented with results like:
- Patient Master
- Visit Narrative Services
- Animal Inventory
- Quarantine
- Animal Patient Ledger
- Treatment Cost by Species
None of these are standard Business Central pages. None of these come from any installed extension. Yet when he clicks on “Treatment Cost by Species,” a fully rendered analysis page appears showing treatment costs broken down by dogs, cats, reptiles — complete with columns for number of treatments, average cost per treatment, and total cost. He drills into the reptile category and gets a detailed reptile treatment cost analysis.
There’s clearly some kind of deep, agentic AI process running behind the scenes — Erik notes a visible processing step each time he clicks on something, suggesting the pages and data are being generated in real time.
Mobster Asset Management and a DMV for Terrible Cars
To really stress-test the system, Erik starts asking for increasingly absurd domains. He types “mobster asset management” and gets results including fixed asset ledgers, asset disposal journals, depreciation books, and fixed asset analysis pages. The depreciation book renders with equipment categories ready to explore.
Then he goes further: “DMV for shitty cars.” The system doesn’t flinch. It returns an inventory valuation report populated with VIN numbers and vehicles. He drills into a 2008 BMW 3 Series and finds:
- Engine problems — runs but overheats
- Rough cost basis and current value with depreciation
- Dented frame, rust around wheel wells
- Automatic transmission that slips occasionally
- Failed emission test
- No AC
Erik muses that depreciation on a DMV record might be for tax purposes — but the point is that the system generated contextually appropriate data and business concepts for a completely made-up scenario.
Extension Management Looks… Different
When Erik navigates to Extension Management, the experience is different from what we know today — it renders as a menu rather than a flat list. The installed extensions have unfamiliar, almost cryptic names: “Ether,” “Data Shield,” “Luminous,” among others. As Erik puts it: “The future’s weird.”
He wisely decides not to touch any of those, wanting to keep the system running for more demos.
Waste Management with Business Logic
The most impressive demo comes when Erik searches for waste management. A scrap journal appears with entries for recycled HDPE pellets, aluminum waste, cardboard, crushed glass, organic compost, and rubber waste. He opens a cardboard waste entry described as “water damaged during storage” and discovers something remarkable: business logic.
The system has determined that contamination was detected, and the disposal method has been automatically updated to incinerate due to the contamination hazard. This isn’t just generated data — it’s generated reasoning. The system applied domain-appropriate rules to the scenario.
As Erik says: “There’s business logic. It’s full-featured. This is so weird.”
A Peek at the Source Code
While Erik doesn’t dive into code in the video, the source files that accompany this build offer some interesting clues. The app.json reveals key details:
{
"name": "Agents",
"publisher": "Default Publisher",
"version": "1.0.0.0",
"application": "26.0.0.0",
"runtime": "15.0",
"idRanges": [
{
"from": 50100,
"to": 50149
}
]
}The app targets application version 26.0 and runtime 15.0 — versions that are well beyond what’s currently publicly available. The app is simply called “Agents”, which aligns with Microsoft’s broader push toward agentic AI across their product line.
The AL source file shows a page extension on the Customer List that references some telling namespaces:
namespace DefaultPublisher.Agents;
using Microsoft.Sales.Customer;
using System.Utilities;
using System.Agents;
pageextension 50100 CustomerListExt extends "Customer List"
{
trigger OnOpenPage();
var
Ref: RecordRef;
TB: Codeunit "Temp Blob";
FR: FieldRef;
InS: InStream;
An: Text;
begin
ClientType::
end;
}The using System.Agents; namespace is the standout here — this is a namespace that doesn’t exist in any current public version of Business Central. The code appears to be a work-in-progress stub, with variables for RecordRef, Temp Blob, FieldRef, and an InStream — suggesting a mechanism that dynamically works with record references and binary data streams, possibly to render AI-generated page content on the fly. The incomplete ClientType:: line hints at client-type detection, perhaps to determine rendering capabilities.
What Does This Mean?
If what Erik is demonstrating is real — and the evidence strongly suggests it is — then Business Central is heading toward a paradigm where:
- Applications are generated on demand rather than pre-built and installed
- Domain-specific functionality doesn’t require custom development — you describe what you need, and the system creates it
- Business logic is inferred from context, not just data structures and UI
- The Tell Me bar becomes an application creation interface, not just a navigation tool
This raises profound questions for the Business Central partner ecosystem. If the platform can generate an animal hospital management system or a waste management solution on the fly, what happens to ISV apps? What happens to custom development projects? The answer likely isn’t that they disappear — real-world business requirements are far more nuanced than what a demo can show — but the baseline of what’s “built in” may shift dramatically.
Conclusion
Erik has stumbled onto something that looks like a genuine glimpse into Business Central’s AI-powered future — a system where the System.Agents namespace and deep agentic AI processing allow the platform to dynamically generate entire business applications, complete with data, pages, and business logic, from nothing more than a natural language prompt. The rendering isn’t perfect yet, and there are clearly rough edges, but the core capability is staggering. As Erik puts it: “I think we just saw the future.” Whether that future is scary or exciting — probably both — stay tuned for his follow-up investigation into exactly what’s powering this experience.
