The Digital Platform principles

Somewhere in your company right now, someone is copying a number from one system into another. An order into the invoicing tool, a customer into the mailing list, a stock count into the spreadsheet where the real planning happens. Every company recognizes this. Almost nobody chose it.

This document is our answer to how that happened and what replaces it. Ten principles, each with its own page that goes deeper into examples, objections and evidence. It starts from the business and works backwards into software, because that is the only direction that produces software worth having.

It is also a living document. As the thinking sharpens and the evidence grows, it changes.

Businesses never wanted software products

Nobody starts a company hoping to one day operate a portfolio of software products. They start it to sell something, build something, serve someone. Yet a few years in, the same inventory has accumulated everywhere:

• CRM
• ERP
• Accounting
• Warehouse
• Production
• HR
• BI
• Ticket systems
• Project management
• Integrations to hold it all together

None of these purchases were wanted for their own sake. Each one stood in for an outcome the business needed, and each was rational on the day it was made. The sum is what nobody would ever have designed: people want one business, not twenty systems.

The patchwork was never a strategy. It was a consequence nobody could avoid.

Legacy was inevitable

Legacy systems are treated as evidence that somebody, somewhere, made bad calls. They are nothing of the sort. Companies bought what existed, consultants recommended what worked, vendors built what could be sold. Everyone acted rationally, and the result was still a mess — because for decades there were only two realistic ways to get software:

Both roads end in the same place: patchwork, complexity, technical debt. That was not poor decision making. It was the only economically viable path, which is exactly why the same pattern repeats in every industry, in every country, at every company age.

Blaming past decisions misses the point. The options were bad, and the options have changed.

What businesses actually want

Businesses do not care about software. They care about outcomes, and software is the price paid to get them.

State the actual want plainly and it is almost embarrassing in its simplicity: the exact workflows, the exact processes, the exact information, the exact automations and the exact interfaces that maximize business value. Nothing more, because everything more is cost and noise. Nothing less, because everything less gets patched with spreadsheets.

A business is one connected system

No department experiences the company as departments. A sale commits production, production consumes material, purchasing spends cash, cash constrains finance, and finance shapes what management decides next. One order touches all of it before lunch.

Software should mirror that reality. Instead, traditional software mirrors vendors: every package draws its border where a product line ends, not where the process does — and the business pays people to carry data across the borders.

The org chart has departments. The business does not.

Software should adapt to the business

For decades the adaptation ran backwards. Companies renamed their own concepts to match the package, rebuilt working processes around release notes, and accepted "that's not how the system works" as a legitimate answer to a business need.

The right direction was always obvious. Software should adapt to the company's workflows, the company's terminology, the company's processes and the company's rules — the company should not redesign itself because its software cannot.

Software should adapt to every individual

Company-level fit is not the end of adaptation. It should not stop at the department, or at the role. It should continue down to each individual, because every employee works differently, thinks differently and needs different information in front of them.

AI makes that economically possible for the first time. Software can reshape itself for every individual while still respecting company policies, permissions and governance — and the individual does the reshaping by describing how they want to work. Not by opening administration panels. Not by writing code. By communicating intent.

What changed

Only now, seven principles in, does technology enter this document — deliberately. AI presented as the story is a pitch. The real shift is the emergence of a different software architecture, of which AI is one part:

• Cloud computing
• APIs
• Composable architectures
• Reusable components
• Modern platforms
• Continuous deployment
• AI-assisted development
• AI-assisted configuration
• AI-assisted workflows

Each of these alone is an increment. Together they broke the economics that made the patchwork inevitable: assembly replaced invention, and change became routine instead of a project.

AI dramatically changes the economics. It does not replace architecture — it enables better architecture.

Custom software no longer means what it used to

"Custom" used to be a warning label: expensive, slow, risky, dependent on whoever built it. That definition described how the software was made — handcrafted from scratch, every time, for every company.

It is obsolete. Modern platforms compose highly customized systems from reusable, proven building blocks; what gets built by hand is the thin layer that is genuinely yours. Custom software is no longer handcrafted. It is assembled intelligently, and the economics moved with the method.

The next generation of business systems

The system this makes possible understands workflows, users, data, business rules and intent. Instead of configuring software, people describe what they want, and the platform adapts.

Everyone works inside the same coherent system and nobody has the same experience of it. Sales gets sales support, production gets production support, management gets decision support — same data, same rules, different mornings.

Measuring success

Software is still bought by counting things: features, modules, integrations, rows in a comparison matrix. Every one of those numbers measures the vendor's output, not the buyer's outcome.

There is one question that matters: how much business value does the system create? Everything else is secondary — not unimportant, secondary. A measurement culture that starts from that question ends procurement theater, and it is a culture we hold ourselves to in public.

Where this ends

For decades businesses accepted fragmented software because there was no alternative. They adapted themselves to their tools, paid people to ferry data across system borders, and called the result IT strategy — because it was the best available move, every single year.

It no longer is. That era is ending.