The next generation.

The lazy AI-era take is that agents will replace business software. The lazier take is that business software is fine and doesn't need to change. Neither view has spent enough time with an actual ERP implementation. Enterprise software is two things, not one — a data layer and an interface layer doing very different jobs. Only one of them is dying.

Business applications are a data layer: the general ledger, the bill of materials, the work order, the dispatch table, the opportunity pipeline, the case queue. The contract between business reality and software. The thing that has to be honest when an agent is wrong. A general ledger that hallucinates is not a feature; it's a lawsuit.

Business applications are also an interface layer: the screens, the workflows, the seventeen-tab forms, the way humans engage with the data. This is where most of the cost lives, where implementations burn twelve months, where users hate the product.

The agentic era takes the second one apart. The data layer holds. The interface layer dissolves, and the interface layer is 80% of what made enterprise software feel old. When people say "ERP is dying," what they mean is "the way I experience ERP is dying." They're right about that. What stays is the system of consistency. What changes is everything else. That's the next generation.

Twenty years ago, ERP looked like a spreadsheet pretending to be a database. Ten years ago, ERP looked like a CRM pretending to be a spreadsheet. Today, ERP looks like a meeting where nobody wants to be.

The screen was the entity that defined ERP for two decades. Now it's the thing the field tech is trying to avoid. So is the project superintendent. So is the dispatcher, the warehouse foreman, the AP clerk. The data has to flow into the system; nobody disagrees about that. They just don't want to be the ones typing it.

The interface becomes conversational, ambient. It becomes a Teams thread, a voice memo to the truck radio, a photo of a damaged pallet that turns into a work order with seven fields auto-populated. The screen doesn't disappear. It just stops being the front door. The front door becomes language.

This is more than a UI redesign. It's the recognition that ERP's interface was never the point. The point was the consistency of the underlying record. Conversational surfaces honor that record without forcing the human to learn the form.

The old ERP was a monolith. You bought the whole thing, you fit your business to its shape, and you spent two years pretending the modules you didn't need were "future phase." The platform was a fortress. The implementation was a siege.

The new ERP is composable. D365 is closer to a set of capabilities than a single product: Finance & Supply Chain, Project Operations, Field Service, Sales, Customer Service, Power Platform, Dataverse. Each is its own surface area, each can be extended, and each can host AI-native modules that didn't exist when the platform was originally designed.

The work, then, is to compose ERP, not to install it. Decide which D365 modules belong in the stack. Which vertical extensions snap on top. Which AI modules sit above the data layer doing things the base product wasn't built to do. The composition is the product. The base is the substrate.

This is why FarShip exists, and why Helios 1 and Helios 2 exist. They're not "alternatives to D365." They're the additive layer that turns a horizontal platform into an operating system for a real business.

The longest line on the average ERP Gantt chart isn't "build," it's "discovery." Six months of workshops where senior business leaders explain how their business works to a delivery team that's never operated it and won't be there when it stops working.

AI breaks this asymmetry. A well-instrumented delivery team can ingest a customer's policy documents, master data, open work orders, and email archives, then arrive at the first design workshop knowing more about how the business runs than the people in the room. Those people still own the decisions; this is preparation, not replacement. It removes the part of discovery where senior leaders pay senior consulting rates to explain things twice.

The same dynamic compresses configuration, testing, and training. AI-assisted configuration writes the first draft. AI-generated test cases cover edges the human team forgets. Training materials, narrated by AI, adapt to each role on the customer team. The project doesn't get cheaper. It gets shorter and lighter, with a measurable 30–40% reduction in both time and effort. That's a different sale.

Our AI Implementation Methodology is the operationalized form of this idea. Every workstream re-engineered with AI as the lever, not the headline. A measurable 30–40% less time and effort, across engagements. The compression isn't a marketing claim. It's the product.

The dominant AI story in enterprise software right now is horizontal: foundation models, agent platforms, generic copilots, "AI for every workflow." That story works for the platforms selling it. It doesn't work for the businesses that have to operate on it.

A horizontal agent doesn't know that a heavy-equipment dealer's parts-aging rules are different from a manufacturer's. It has no idea that an EPC project codes labor differently in week 8 of a billable phase than in week 2. It can't even tell you what a CAT 980G is, much less how its lifecycle should be modeled in the asset hierarchy.

The depth lives in the vertical. The vertical is where the data is messy enough, the workflows are specific enough, and the cost of being wrong is high enough that generic AI breaks down and a specific AI is worth real money. Heavy equipment. Infrastructure-build E&C. Energy. Industrial professional services. These aren't "verticals" in a marketing-pages sense; they're the actual places where ERP either earns its keep or doesn't.

Our products go vertical because the value is vertical. FarShip ships D365 into heavy equipment dealer networks because heavy equipment is a real industry with real characteristics, not a slide in a deck. Helios 2 ships configured for the customer's specific industry because the generic version is the one that breaks. We'd rather be indispensable in three industries than tolerated in thirty.

Four things, on the same thesis.

FarShip — a vertical D365 F&SC extension for heavy equipment, infrastructure-build, and dealer networks. Eight weeks to a customer-ready model, every dealer / territory / line of equipment modeled once.

AI Implementation Methodology — the operationalized form of the velocity-compression argument. A measurable 30–40% reduction in time and effort across the implementation. The methodology already runs in production. Every new Ludia engagement uses it as the default scaffold.

Helios 1 — Ludia's own AI operating system. Every workstream of the firm is mapped: methodology, pipeline, delivery patterns, customer data. The whole thing runs on top of an AI layer. Our team uses it every day. We have mapped our entire business and built it ready for agility. In production.

Helios 2 — the customer version of Helios 1. Same architecture and AI core, configured for your industry, your data, your operating patterns. The way we run Ludia, productized for the way you run your business. Available now and ready to help on your journey.

If you're a CIO, COO, or CFO at a $300M–$1B industrial firm and any of this matches a problem you're losing sleep over, that's the conversation we want to have. Not a demo. Not a pitch. A two-hour, two-side, what-are-we-actually-solving-for conversation. We'll come prepared.