DRT Week 10A. Deterritorialization / Reterritorialization as innovation sequencing
Escape that must become livable
Three short puzzles
Puzzle 1 — The “agentic fabric” that doesn’t stick (Infosys).
Infosys can credibly sell an AI-first stack—Topaz, Topaz Fabric, and an agentic foundry logic that promises reusable, governed agents across client processes and internal operations. Yet what often determines whether “agentic” becomes a durable operating advantage is not model quality, but whether new roles, boundaries, and accountability patterns become routine rather than special. When humans are nominally “in/on/off the loop,” the unglamorous question is: who owns the loop when something goes strange at scale?
Puzzle 2 — Great analytics, uneven reality (ITC).
ITC’s AI-enabled sensing and execution is impressive precisely because it spans heterogeneous contexts: social listening and research acceleration, agribusiness procurement, manufacturing quality, retail execution, and B2B platforms. Yet the hardest friction shows up where “the same” system meets different field realities: different outlets, regions, crops, credit constraints, and capabilities. The tool works—until it meets a local exception that becomes a permanent workaround.
Puzzle 3 — Digital twins everywhere, discipline nowhere (Siemens).
Siemens’ portfolio—automation, digital twins, copilots, edge AI—points toward a coherent industrial future. The paradox is that industrial transformation fails less from lack of technology than from lack of stabilization across layers: engineering, operations, safety, cybersecurity, and governance. If the organization cannot stabilize what the twin means for decisions on the shop floor, the twin becomes theater.
Across all three puzzles, novelty is not the bottleneck. Making novelty inhabitable is.
The hidden common structure
Each puzzle is a two-move dynamic that organizations routinely confuse as “innovation”:
Move 1: Escape (deterritorialization).
A break from established routines and couplings. New experiments, new toolchains, new cross-functional linkages, new ways of “seeing” the system. The organization becomes mobile—sometimes exhilaratingly so.Move 2: Inhabitation (reterritorialization).
The slow construction of supports that make the new livable: decision rights, interfaces with embedded constraints, training and credentialing, auditability, repair loops, rollback readiness, and social legitimacy.
The recurring failure is treating Move 1 as “the work,” and Move 2 as “bureaucracy.”
That is how pilot portfolios become churn portfolios.
Concept reveal and definition
Deterritorialization is organized escape: the deliberate loosening of an existing configuration so that new relations can form—new experiments, new couplings, new ways of composing work. It raises the system’s mobility.
Reterritorialization is organized inhabitation: the deliberate building of supports so that what was discovered in escape can be enacted as a governable practice—repeatable, legible, auditable, repairable, and teachable.
In DRT terms (kept managerially usable): the sequence is not “idea → rollout.” It is a cycle across the MLXE domains:
L (virtual-real life territories): what becomes real for people—habits, trust, identity, and the felt sense of “how work works.”
E (actual-real flows): what becomes operationally real—signals, workflows, handoffs, performance, failure modes.
X (virtual-possible experience universes): what becomes thinkable and sayable—narratives, meanings, scenarios, and what “counts” as a good outcome.
M (actual-possible machinic options): what becomes structurally possible to enact—platform capabilities, constraints, and authorized change pathways.
Deterritorialization tends to open L and X (new lived possibilities and new interpretive frames) by perturbing E and M (existing flows and existing structures).
Reterritorialization closes the loop by stabilizing E and M without killing L and X.
You are not innovating if you cannot make the new inhabitable across L (lived), E (operational), X (meaningful), and M (structurally enactable).
How it rewires each puzzle
Infosys, re-seen:
Topaz Fabric and agentic workflows are deterritorializing forces because they recompose who does what, how exceptions are handled, and where judgment sits. The success condition is not “agent deployed,” but agent governance made ordinary. That requires new micro-rituals (how supervision happens), new interfaces (constraints embedded into workflows), and explicit rollback and escalation pathways. Otherwise, humans remain “in the loop” only as informal firefighters—high cost, low learning.
ITC, re-seen:
When AI spans farm-to-kirana, reterritorialization is not a central template; it is a portfolio of local stabilizations anchored to shared principles. What must stabilize is: (i) what evidence counts in different contexts, (ii) how exceptions are handled without informalism, and (iii) how learning travels from one region/outlet/crop context to another without becoming brittle. Without that, adoption issues are misdiagnosed as “change resistance” rather than “insufficient inhabitation.”
Siemens, re-seen:
Industrial transformation is a textbook case of disciplined reterritorialization. Digital twins, copilots, and edge AI generate new couplings across engineering and operations. But safety-critical environments require that accountability and traceability are not afterthoughts. Reterritorialization here means: interfaces that prevent unsafe actions, auditable decision pathways, and training that changes competence—not just tool familiarity. The prize is not “more AI,” but more reliable agency across the system.
Bonus: Catena-X
Standardization is reterritorialization in its purest form: shared semantics, shared obligations, and shared constraints that make cross-organization coordination possible. In ecosystems, “escape” is easy (everyone builds their own portal, their own data model, their own agent). Inhabitation is hard (shared meaning, shared rules, shared trust mechanisms). Catena-X’s logic is precisely that: interoperability and sovereignty are not slogans; they are stabilization work.
Where it breaks
This lens misleads in two situations:
When the environment is genuinely too volatile to stabilize.
If the regulatory, technological, or geopolitical substrate is shifting weekly, heavy reterritorialization can become premature hardening.When the organization confuses stability with centralization.
Reterritorialization is not synonymous with “control by headquarters.” It can (and often must) be distributed: common principles, local adaptation, and strong repair loops.
The test is simple: does stabilization increase learning velocity and trust, or does it freeze the system?
How to recognize when to use it (signals + diagnostic questions)
Signals you are in a deterritorialization/reterritorialization problem:
Pilot success but rollout fragility.
“Shadow practices” proliferate (spreadsheets, side channels, informal approvals).
Human-in-the-loop exists mostly as heroics.
The same tool behaves differently across sites, teams, or clients.
Diagnostic questions (use in a steering committee this week):
What exactly are we deterritorializing? (roles, workflows, data regimes, customer interfaces, decision authority)
What must become stable for this to be safe and repeatable? (interfaces, constraints, auditability, training, escalation)
Where will exceptions live—formally? (authorized change classes, rollback readiness, repair loops)
What is the smallest “inhabitable unit” of the new practice across L/E/X/M?
What are we willing to keep exploratory—and for how long?