AI’s Big Payoff Is Coordination, Not Automation

AI’s economic impact is usually explained in terms of falling costs of prediction or falling costs of creation. But there’s a more important factor that may be the least understood and most significant: falling costs of translation. Translation sounds trivial until you notice its role in modern work: the constant effort of turning one team’s outputs into another team’s inputs. Translation costs show up in the effort required to convert or reconcile outputs, or the time lost in meetings and rework. When translation works, it enables a disjointed set of actors, tools, and data to work together toward a common outcome—it enables coordination, in other words, allowing projects to move forward reliably instead of getting stuck in reconciliation and rework. AI makes translation cheap and general. It does this in two ways: It extracts structure from unstructured information, making it usable, and it then employs this usable information to get work done. Complements that once failed to fit together, because their data, vocabularies, or workflows did not match, can now be easily combined without forcing agreement on tools or workflows. For the first time, coordination becomes possible without consensus, which means it can be extended to professional arenas where it was previously too difficult or costly. Over the past decade, my work has focused on designing platforms to organize and coordinate fragmented systems. But until recently, that kind of coordination had a hard limit: It depended on highly structured data and clean interfaces, which required consensus on standards and workflows. In practice, most systems varied too widely to align, so coordination remained manual and expensive. But AI’s translation capabilities have removed that limit. The new kinds of coordination that it enables are changing not just workflows but also the nature of value creation and competition in fragmented industries. Coordination without Consensus Consider how this plays out in the architecture and construction industry. Most construction projects require that multiple specialized professionals—architects, structural engineers, contractors, and others—work together. But each group has its own set of special tools and concerns. Architects care about spatial relationships and aesthetics. Structural engineers focus on load paths and safety factors. Contractors prioritize sequencing and project management. Coordination in this context becomes mission-critical. Construction projects frequently deviate from original specifications, and if different teams don’t coordinate effectively to adapt, the results are predictable: rework, delays, and performance issues in finished buildings. Attempts to solve this problem through top-down standardization—forcing everyone involved in the project onto a unified platform or common data model—have repeatedly failed or enabled limited coordination in one activity while throwing another out of alignment. Additionally, and not surprisingly, specialized teams resist abandoning the tools that define their craft. Building designs created with one tool do not easily interoperate with construction plans being managed in another, and if they’re not aligned, you get high translation costs. AI offers a new kind of translation that can solve these problems. Instead of requiring every team to adopt the same tools, formats, or standards, it takes fragmentation for granted and pulls information from sources as disparate as specialized BIM software, spreadsheets, photos, emails, and annotated PDFs, each of which only captures a partial view of the project. This integration creates a unified view of the project, making it possible at any moment for anybody involved to see the live progression of work across design, engineering, scheduling, and compliance. With this kind of unified view, project managers can oversee a fragmented process while allowing each team to preserve its own ways of working. They can also make decisions that resolve trade-offs across disciplines without needing to convene endless coordination meetings. If the architect moves a stairwell, the AI layer can identify downstream effects on structural beams and notify the structural engineer. AI becomes the hidden coordination layer that continuously checks and reconciles changes across teams, synchronizing execution through continuous translation. Sensing opportunity in the industry, many companies are getting into the game. Trunk Tools, for example, gathers information from tools such as Autodesk (used by architects and engineers to create designs and plans) and Procore (used by contractors to manage their jobs). After reading through specifications, drawings, and schedules, Trunk Tools organizes a structured, searchable project record, which means that architects, engineers, and contractors don’t need to waste time comparing notes with one another and resolving misalignments. Instead, they can direct all questions to a single source and receive answers grounded in the relevant documents. Incumbents are sensing opportunity here too, as illustrated by Procore’s acquisition of Datagrid, which takes a similar approach to extracting structure from unstructured files and sources. A New Kind of Battle When coordination no longer requires consensus, firms that built their dominance on enforcing common standards, interfaces, and workflows begin to lose the foundation of their advantage. Consider a case in the U.S. auto-insurance market,where the startup Tractable is challenging the dominant incumbent CCC Intelli gent Solutions, despite CCC’s entrenched advantages. When you make an auto claim, it moves through a web of independent actors (insurers, repair shops, parts suppliers, appraisers), each of which uses different software systems and works under different incentives. CCC took control of its market by establishing a shared language of standardized damage codes and prescribed digital workflows—a language that has been adopted by the various parties involved in handling claims. For a competitor to unseat CCC, it would have to replace CCC’s standards, but that, in turn, would require abandoning today’s tools, retraining adjusters, and rewriting insurer approval processes. The switching costs have long just seemed too high, and for the past two decades CCC’s dominance has appeared secure. But no longer. In its bid to unseat CCC, Tractable decided to sidestep the standards battle entirely. Instead of attempting to persuade insurers and repair shops to abandon CCC’s codes, it trained AI models to interpret unstructured smartphone photos of damaged vehicles and produce repair estimates that could be fed directly into existing insurer workflows. Tractable enabled coordination without consensus, in other words, and the results are impressive: By 2023, it was processing nearly $7 billion worth of claims on behalf of major insurers. The Way Forward The opportunity to leverage AI’s translation capabilities extends far beyond the industries discussed in this article. In any fragmented industry, from healthcare to logistics, AI’s ability to reconstruct a shared state across disconnected systems creates both opportunity and threat. Incumbents have three possible strategies available to them. 1) Become the translation layer. Firms that choose this route accept that proprietary standards will no longer be sufficient to secure their position. Instead, they will invest in reconstructing the shared view of ecosystem activity more effectively than anyone else. In logistics, platforms like project44 have taken this approach, creating a unified view of shipments without requiring carriers to adopt new standards. Companies pursuing this strategy pursue openness across tools and formats to make themselves indispensable as the interpreter of ecosystem activity. 2) Double down on accountability. Instead of competing to become a neutral translation platform, incumbents can emphasize their ability to assume responsibility for outcomes, even as others translate. Maersk exemplifies this strategy: The company has moved toward integrated logistics, expanding into end-to-end contracts, customs brokerage, warehousing, digital booking systems, and risk-managed execution. In a world of supply-chain disruption, sanctions risk, and regulatory scrutiny, Maersk’s integrated control allows it to guarantee execution and manage risk better than others, and compete on that basis. 3) Fragment and tax. Here incumbents neither fully embrace open translation nor retreat entirely into governance and accountability. Instead, they use AI-aided translation to consolidate a privileged, unified view inside their own walls—then ration access to that unified view to the wider partner ecosystem which remains fragmented. FedEx has adopted this strategy. In recent years, the company has invested heavily in AI-driven routing systems that rely on a unified, real-time picture of where shipments are and what condition they’re in across its entire network. However, it exposes only slices of that unified picture externally—and often on paid terms—so partners and customers must rely on FedEx to access complete and timely information. This approach allows FedEx to capture the benefits of translation inside its own network while ensuring that others remain dependent on its infrastructure, and it will remain viable as long as FedEx’s network density and service coverage keep customers willing to pay for visibility rather than reconstruct it themselves. . . . As AI-aided translation enables different people, tools, and systems to work together without consensus, it will unlock real gains in speed, cost, and innovation across fragmented industries, but it will also shift the locus of power. In the short run, the winners will be those who use AI to cut through friction and get work done despite mismatched systems. Over time, however, coordination alone won’t be enough: When projects scale and failures carry real consequences, questions of responsibility, trust, and liability will force a return to shared rules and clear accountability. The practical path forward is to use AI to jump-start collaboration where coordination has long been too costly, while deliberately building the governance, contracts, and norms that turn fast coordination into durable, trustworthy ecosystems.

AI’s Big Payoff Is Coordination, Not Automation

Summary.

Partner Center