RESHUFFLE An interactive companion to the book
Explainer · Chapter 3 / 11 ▸ Frames cluster

AI coordinates fragmented systems without making them agree first.

Every coordination breakthrough until now required upfront standardisation. Container, barcode, ISO formats - everyone had to adopt the same thing first. AI is the first coordination technology that doesn't.

Most coordination plays in the last century started the same way: a long, painful standardisation phase. Get everyone to adopt the same protocol. Then coordinate on top of it.

AI breaks that pattern. It reads whatever format you bring. Which means coordination can now happen in places it previously couldn't reach.

▍ The mechanism

The standardisation tax goes away

Container shipping took twenty years to standardise - every port, every shipping line, every railway had to agree on box dimensions and a single bill of lading. The barcode took a decade across retail. EDI took longer. Every coordination breakthrough was bottlenecked by the politics and economics of getting everyone to agree on a format.

AI dissolves the dependency. It reads unstructured smartphone photos, free-text descriptions, voice memos, PDFs in different layouts - and produces a shared representation that everyone downstream can act on. The consensus layer isn't gone; it's implicit, learned from data rather than negotiated upfront.

That changes the economics of who can lead coordination. Previously: only large institutions with negotiating power (governments, retailers like Walmart, standards bodies) could force consensus. Now: any firm with sufficient AI capability and training data can stand up a coordination layer in an industry that's never been coordinated before. The barrier to coordination collapsed. So did the kind of firm that could lead it.

▍ Historical analogue

CCC vs Tractable in auto-insurance claims

The same control point, won two ways.

CCC Intelligent Solutions coordinates the US auto-insurance claims ecosystem through consensus. Every dent and scratch gets tagged using CCC's predefined damage codes. Every insurer, repair shop, and parts supplier uses CCC's shared vocabulary. The system runs smoothly because everyone structures their operations around the common standard.

Tractable, a UK competitor, won the same control point differently. Their AI reads unstructured smartphone photos of damaged cars and generates repair estimates directly - no codes, no taxonomies, no upfront agreement. The damage model emerges from millions of past claims, learned by the AI. Insurers and shops keep using their existing tools. The coordination happens silently in the translation layer.

Same control point. Two completely different ways of getting there. The CCC playbook took decades; Tractable assembled comparable coordination in a fraction of the time. The shift isn't only about who wins - it's about how many industries are now coordinatable that weren't before.

▍ Two coordination architectures

Consensus-based vs AI-translated

Architecture A
Consensus coordination
Setup
Everyone adopts the same format first
Speed
Slow - years to roll out the standard
Reach
Limited to actors who can/will adopt the standard
Examples
Container shipping, barcode, EDI, CCC
Defence
Switching costs, embedded protocols
Architecture B
AI-translated coordination
Setup
Partners keep their existing formats; AI translates
Speed
Fast - no standardisation phase
Reach
Anywhere AI can read the inputs
Examples
Tractable, modern voice agents, multimodal AI
Defence
Training data scale + compounding model accuracy

The old architecture took decades to build and was rigid once built. The new architecture builds in months and adapts as the ecosystem changes. The strategic question is which architecture your industry's next coordination layer will use.

▍ Where this matters most

The industries that have stayed uncoordinatable until now

01

Fragmented professional services

Legal, accounting, healthcare - industries where every actor uses different forms, different terminology, different systems. Decades of EDI and standardisation pushes have failed to consolidate them.

AI translates across the chaos. The firm that builds the coordination layer first owns a position that standardisation pushes have failed to win for half a century.

02

SMB and long-tail B2B

Small and medium businesses use thousands of different systems. Forcing them onto a common platform is what every B2B SaaS has tried and mostly failed at.

AI doesn't need them to adopt anything new. It reads invoices in any format, voice memos, screenshots, spreadsheets - and coordinates across what they're already doing. Whole categories of B2B coordination become buildable that weren't before.

▍ Apply it

Two ways to win the same coordination layer

Pick a path. The next decade plays out beneath your choice.

You need to coordinate fragmented actors in a new ecosystem. You can build the system on consensus (force everyone to agree on standards first) or without it (let AI translate across formats). Each path runs differently.