Dynamic Capabilities

Confidence 0.95 · 50 sources · last confirmed 2026-10-15

A firm-level capability for sensing opportunities and threats, seizing them, and transforming the firm’s resource base in response to changing environments. Distinguished from ordinary capabilities (doing things right; replicable; outsourceable) by their role in governing the rate of change of ordinary capabilities. Origin: David Teece (1997, 2007).

Working definition

Per Teece (2007, as quoted in Warner & Wäger 2019):

Dynamic capabilities = “a company’s capacity to (a) sense and shape opportunities and threats, (b) seize opportunities, and (c) maintain competitiveness through enhancing, combining, protecting, and, when necessary, reconfiguring the business enterprise’s intangible and tangible assets.”

“Dynamic capabilities are about doing the right things, whereas ordinary capabilities are about doing things right.” (Teece & Leih 2016)

Key claims

The three-cluster framework (Teece 2007)

Cluster	Function
Sensing	Scanning the external environment for trends/threats; opportunity identification
Seizing	Mobilizing resources to capture opportunities; new business model design
Transforming	Reconfiguring the firm’s asset base; renewal of structures and culture

Microfoundations for digital transformation (Warner & Wäger 2019)

Empirically identified across 7 incumbent German MNCs and 18 strategy-consultant interviews — nine subcapabilities organized under the three clusters:

Cluster	Microfoundation	What it does
Digital Sensing	Digital scouting	Tech trends; competitor screening; customer-centric trend sensing
	Digital scenario planning	Signal analysis; future-scenario interpretation; digital-strategy formulation
	Digital mindset crafting	Long-term vision; entrepreneurial mindset; cultural promotion
Digital Seizing	Rapid prototyping	MVPs; lean startup; digital innovation lab
	Balancing digital portfolios	Internal/external option balance; scaling new BMs
	Strategic agility	Rapid resource reallocation; redirection acceptance; strategic pacing
Digital Transforming	Navigating innovation ecosystems	Partner interaction; co-creation; ecosystem capabilities
	Redesigning internal structures	CDO appointment; team-based structures; BM digitalization
	Improving digital maturity	Workforce maturity; digital natives; internal knowledge leverage

Three forms of strategic renewal that result

• Business model renewal — replacing transactional product logics with relational/multi-sided value propositions.
• Collaborative-approach renewal — replacing siloed, internal-only collaboration with cross-functional and external-ecosystem collaboration.
• Cultural renewal — refreshing or replacing legacy cultures with digital-mindset / entrepreneurial cultures.

Contextual factors

External triggers	Internal enablers	Internal barriers
Disruptive digital competitors	Cross-functional teams	Rigid strategic planning
Changing consumer behaviors	Fast decision making	Change resistances
Disruptive digital technologies	Executive support	High level of hierarchy

Why digital transformation requires new dynamic capabilities

• New digital technologies (AI, cloud, IoT, blockchain) change the nature and purpose of dynamic capabilities — not merely their content.
• Organizations can now scale up/down at speed, ease, and cost not previously possible.
• The convergence and generativity of digital technologies forces incumbents to behave entrepreneurially even when entering competitively established markets.

Operator-narrated cases at mid-tier regional incumbent scale (DFI 2026)

Scott Price’s CNBC Managing Asia interview (May 2026) supplies a compact case set of seizing- and transforming-cluster microfoundations operating at multi-brand multi-country retail-incumbent scale (DFI Retail Group: ~thousands of supermarket / 7-Eleven / Guardian / Mannings / IKEA-Asia / Maxim’s outlets across HK / SE Asia / mainland China). All cases are first-person CEO-narrated and have explicit named-numbers anchors:

Microfoundation	DFI case
Balancing digital portfolios (seizing)	Sold the Singapore supermarket business for S$125M (~US$93M); closed all ~100 Mannings stores in mainland China after concluding 2,000 stores would be needed to win at scale, retaining only online presence via Chinese e-commerce platforms; redirected capital to Southeast Asia health & beauty (now the strongest segment, 1,500+ stores).
Strategic agility (seizing)	“We source from more than 50 countries around the world. We always have to have the ability to pivot very quickly to protect that pricing to customers.” — supply-chain pivot capability named explicitly.
Navigating innovation ecosystems (transforming)	The Yuu loyalty platform as a data-monetization flywheel linking millions of shoppers across DFI’s brand portfolio — “the way you protect the bottom line for shareholders is you create your own digital revenue that has a higher margin. Data is the core to that.” Vendor-insight sales + cross-segment promotion-permissioning already monetised.
Business model renewal (strategic renewal)	The Chinese Wellness Hub at Mannings HK — TCM-practitioner consultation + in-store health pod with basic-vitals measurement; pivot from commodity shampoo retail to functional-wellness platform. Plus the low-water rice programme in Thailand as a scope-3-emissions / value-pricing renewal (sold in stores at the same price — “our customers won’t pay a penny more”).
External-triggers sensing (contextual)	Named the agentic-AI personal-assistant disintermediation thesis as “what keeps me up at night” — the seller-side mirror of Ognibeni’s buyer-side warning that search-driven e-commerce will be the first format agents kill.

The 2,000-store competitive-scale floor is particularly reusable as a Western/regional-incumbent-anchored data point for the balancing-digital-portfolios microfoundation — a public CEO articulation of what scale is required to win against established Chinese platform incumbents in their home market, a quantification the W&W literature treats only abstractly.

End-to-end practitioner operationalisation of the W&W process model at AWS-vendor altitude ( AWS London Exec Forum 2026)

Where the DFI case set anchors specific microfoundations with named-numbers worked cases at mid-tier retail-incumbent scale, Allen traces the entire W&W process arc end-to-end at AWS-Executive-in-Residence advisory altitude:

W&W bucket	Allen’s operationalisation
Digital sensing	Anthropic labour-market report + MIT NANDA 95%-of-AI-pilots-fail framing + Nvidia SLM paper + Jevons-paradox / Schumpeterian-disruption macro-frame as the digital-scenario-planning discipline.
Digital seizing	The USE / COMPOSE / BUILD economic-decision framework as balancing-digital-portfolios; Brooklyn Solutions’ 4-phase iterative progression (basic → conversational → agentic → multi-agent) as rapid-prototyping; the embedded-pod model as strategic-agility.
Digital transforming	The hourglass-organization shape + builders-to-orchestrators role-shift as redesigning-internal-structures; data-engineers-as-context-architects as improving-digital-maturity.
Strategic renewal	The moats-erosion thesis as the load-bearing business-model strategic-renewal claim: the old moats (workflow embeddedness, software scale, integration lock-in, engineering complexity, IP) erode under agentic AI; replacement moats — compounding proprietary data, network effects, regulatory permission, capital at scale, physical infrastructure, time that can’t be parallelised — re-anchor sustainable competitive advantage. The bank-branch-network expansion worked example operationalises physical infrastructure as moat under the new conditions.
Contextual	The junior-hiring crisis (Ravio 73% European-tech entry-collapse) as external-trigger; AWS’s CFO-office partnership for opportunity-cost measurement as internal-enabler; toll-gate / ticket-culture legacy enterprise discipline as internal-barrier.

This is the wiki’s first vendor-altitude end-to-end W&W operationalisation source — distinct from the theoretical anchor itself, from the DFI single-incumbent case set, and from advisory-firm operationalisations like McKinsey. Allen’s keynote is best read as the AWS-advisory-channel translation of the W&W process model into agentic-AI-era enterprise prescription.

Non-AI control case — industrial transformation at Rolls-Royce (Erginbilgiç 2026)

Erginbilgiç’s Rolls-Royce turnaround supplies the wiki’s first pure non-AI industrial-transformation anchor for the dynamic-capabilities lens — every prior source tagged with strategic-renewal/* cells has been AI-adoption-flavoured. The non-AI case is load-bearing because it allows the wiki to separate what’s specific to AI-era dynamic capabilities from what’s dynamic-capabilities primitives full stop.

Mapped to the Teece sense / seize / transform clusters:

Teece cluster	Erginbilgiç’s operationalisation (Rolls-Royce 2023–2026)
Sensing	External benchmarking commissioned Sept ‘22 before the Jan ‘23 start date — “put the mirror up for the organisation. You cannot say the things you just said without data” (~2:01–2:32). Resilience-as-scenario-rehearsal (~21:14–21:32): “It’s not about actually predicting the world, it is about how your company now thinks about dealing with external shocks.” Sensing here is organisational habit of dealing with shocks, not forecasting accuracy.
Seizing	The Jan ‘23 burning-platform speech as the speed-of-commitment moment + the four-pillars framework (people + granular strategy + commercial discipline + performance culture, ~11:03–18:42, with two pillars explicitly named) as the resource-reallocation logic. CEO-to-CEO contract renegotiation (~13:46–15:44) as direct seizing of margin restructuring.
Transforming	Layer elimination without operational-people cuts (~4:24–5:22); “we eliminated layers in the organisation… no operational people left” — transforming the org structure without losing institutional capability. The new-normal-as-eased-cadence observation (~7:36–8:23) is the transformation-as-completed signal: leader demand-intensity drops because team behaviour shifts to new norms.

The convergence with the AI-era anchors is more informative than any single case:

• Strategic-renewal/organizational-culture as the load-bearing W&W cell holds outside the digital lens. Erginbilgiç’s culture-refresh (“non-compromising mediocrity at that level kills the organisations”, ~17:42–17:57) is structurally identical to what Allen 2026’s AWS Executive Forum names at the demanding-leader-as-cadence-primitive layer — different domains, same pattern.
• Strategic-agility as a process capability (the company’s habit) rather than a forecasting capability — Erginbilgiç articulates this explicitly; Krakowski 2025’s tailored-augmentation effects are the AI-era expression of the same primitive.
• The McKinsey-named “case study in the art of corporate transformation” validator anchors the case at consulting-firm altitude without being a McKinsey publication itself — third-party validation of the underlying transformation mechanics.

The implication for the dynamic-capabilities concept: the AI-era literature (Warner & Wäger 2019, the W&W-process-model operationalisations above) is best read as digital-flavoured variations on transformation primitives that the non-AI literature has been articulating for decades. The non-AI control case is a useful corrective against over-attributing the mechanics to AI-specific causes.

The richest operator-altitude case — DBS Bank’s decade-long innovation system ( DBS 2026)

Bidyut Dumra’s MIT SMR Leaders at All Levels interview supplies the wiki’s most complete single-source operationalisation of the Teece sense → seize → transform arc — all three clusters narrated first-person by the executive (Group Head of Innovation and Future of Work) who owns the system, at 39,000-employee banking-incumbent scale across a decade (2009 → 2026). Where DFI anchors specific microfoundations and AWS traces the arc at vendor-advisory altitude, DBS supplies the lived end-to-end case:

Teece cluster	DBS operationalisation
Sensing	The 2014 environmental scan (fintech flurry + Google Play / Apple Pay) → the GANDALF re-framing (competition is big tech, not banks; “be the D in GANDALF”). Competitor re-framing as a sensing act, anchoring the “best bank in the world by 2020” yardstick.
Seizing	The Innovation Pyramid (big bets / Horizon 3 / journeys / entrepreneurs) as portfolio-balancing; the QPR + slush fund as strategic-agility rituals (mid-cycle reprioritisation, “the funding follows suit”); 48-hour build sprints + agent-building as rapid-prototyping.
Transforming	Managing Through Journeys — reorienting the operating model horizontally around customer intent (“a customer is beyond a process — it’s an intent”), mini-CEO leadership, changed incentive/review structures; the 20%-of-scorecard transformation KPI + central-transformation-team playbook as improving-digital-maturity.
Strategic renewal	The “AI-enabled bank with a heart” value-proposition renewal + innovation-is-not-a-choice culture (“don’t tone it down, turn it up”).

Two reusable primitives the DBS case sharpens: (a) innovation-as-KPI — “all parts of the organization have a KPI” — the mechanism that converts a transformation aspiration into a measured org-wide obligation, structurally identical to Erginbilgiç’s performance-culture pillar but in a digital/AI-flavoured incumbent; (b) governance flex for genuine novelty — Horizon-3 bets launch without a business case (written retrospectively a year later) because “if I can write a business case and I know exactly what’s going to happen, I’m not really pushing the needle” — a concrete operationalisation of the balancing-internal-and-external-options microfoundation under uncertainty. The DBS case completes the wiki’s operator/vendor/CEO-non-AI triangulation of the concept with a fourth corner: operator-altitude, AI-flavoured, decade-long, banking incumbent.

Advisory-altitude AI-era read ( LangChain Interrupt 2026)

Andrew Ng supplies a fifth altitude — the advisor-to-the-G2000 vantage (via AI Aspire) — and frames the AI-era version of the sense/seize/transform loop crisply: sensing as continuous scanning of the coding-agent and vendor frontier; seizing as strategic-agility through optionality (≤1-year contracts, open-weight hedging, vendor-neutral observability) and portfolio-balancing (narrowing 300-idea spreadsheets to a handful of high-conviction bets, swing-for-the-fences over incremental); transforming as redesigning the whole workflow (the 10-minute-loan example) via small high-context generalist teams plus the data-architecture rework needed to feed agents. His central claim — bottom-up “thousand flowers” innovation generates point solutions; the transformation needs a complementary top-down motion to redesign the workflow — is a clean restatement of why dynamic capabilities are a system (sensing + seizing + transforming together), not a pile of point solutions. See enterprise-ai-adoption for the full treatment.

The clearest quantified case for portfolio-balancing under uncertainty — corporate venture building at McKinsey scale ( McKinsey Podcast 2026)

Where every prior case in this section supplies a single-firm or single-advisor narration of digital-seizing/balancing-digital-portfolios, Jason Bello’s McKinsey research is the wiki’s first cross-firm, quantified claim about the microfoundation itself: companies building three or more ventures simultaneously dramatically outperform those that try once, and the average cost to reach break-even on a new venture fell from ~$125M (2024) to ~$77M (2025) — a measured trend, not a single narrated case.

Two mechanisms sharpen the microfoundation further:

• Milestone-tranche funding as the seizing-cluster funding primitive. Splitting a venture’s total investment horizon into milestone-gated chunks (e.g. three 3-month tranches within a 9-month goal), releasing the next tranche only after checking whether the milestone was met, is a portable, generalized version of the mechanism DBS’s QPR + slush fund system operationalises at one specific bank.
• Fact-based, blame-free culture as the condition that makes fast pivots possible. “If the facts tell us our product stinks, so be it… there’s no fingerpointing” is Bello’s articulation of strategic-renewal/organizational-culture — structurally the same claim Carroll’s academic culture-as-social-control-system theory predicts, and the same pattern Erginbilgiç’s non-digital performance culture exhibits.

Bello also names the corporate-vs-startup asymmetry directly: corporate venture builders spend far less time fundraising (the board and leadership already know them) and start with incumbent assets (existing customer base, untapped proprietary data/IP) that independent founders must build from zero — a structural explanation for why the portfolio-balancing microfoundation is more affordable for incumbents than for standalone startups attempting the same “multiple shots on goal” strategy.

Small-team, real-time redesigning-internal-structures — a practitioner’s own lived case ( AI Native DevCon 2026)

Where DBS supplies digital-transforming/redesigning-internal-structures at 39,000-employee banking-incumbent scale over a decade, Hannah Foxwell’s talk supplies the small-team, real-time instance: her own 2-person startup ran out of planned work by lunch on day one, forcing an immediate structural response (relearning ruthless prioritization, thinking further ahead, and — echoing DBS’s mini-CEO pattern at radically smaller scale — most subsequent time going to platform engineering and reliability rather than feature work).

Two mechanisms sharpen the microfoundation:

• Team-ratio experimentation as the redesign lever. Some teams trying 2 developers : 1 PM (vs. the traditional 6-8 : 1); Andrew Ng reportedly proposing the inverse at Davos (2 PMs : 1 developer), on the logic that decision-making speed, not coding speed, now bounds a single developer’s usable backlog. New role patterns — the vibe-coding product manager, the forward-deployed engineer, the product engineer — all reduce the same structural distance between the person who understands the problem and the artifact that solves it.
• “Minimum viable human” as a structural floor. An agent can’t hold an on-call pager — sustainable rotation (no one on-call more than 50% of the time, always primary + secondary) sets a hard lower bound on team size independent of how much coding velocity agents supply. A concrete, quantifiable counter-weight to unbounded headcount-reduction narratives.

Foxwell’s talk also touches strategic-renewal/organizational-culture directly — questioning mandatory code review as unsustainable at AI-authored-code volume, and citing Sophie Weston’s “broken comb” (not T-shaped) framing for what career depth should look like when a single generalist now owns more surface area.

Redesigning-internal-structures + rapid-prototyping at a resource-constrained nonprofit ( Silicon Valley Girl, July 2026)

Khan Academy supplies a nonprofit-sector instance of digital-transforming/redesigning-internal-structures, distinct in kind from the professional-services (Priest/PwC), banking-incumbent (Dumra/DBS), and startup (Foxwell) instances already on this page: designers and product managers are given full development environments — pull requests, deployment — explicitly blending role boundaries the organization inherited from “the late 90s into the early 2000s… how do you build web apps” model. CEO Sal Khan frames this as deliberate, not incidental: engineering, design, and product management are “really kind of blending” and the organization’s formal career rubric now requires “learning new tools and adapting.”

The same source supplies a digital-seizing/rapid-prototyping instance: an engineer vibe-coded a game-like feature against real production data at a hackathon; it shipped in about a month, versus a prior estimate of “maybe next school year.” Khan explicitly names the underlying seizing-capability shift: “before, people have said engineering is hard, so do as much time as possible on design… now… let’s build stuff fast, put it out there” — a live articulation of the rapid-prototyping microfoundation’s MVP/lean-startup logic, bounded (per Khan) to lower-stakes, front-end-only features where a fast, cheap failure is tolerable.

The source’s contextual/external-triggers tag — a VC portfolio company automating ~80% of a Philippines-based call-center workforce — is the shock that motivated Khan’s broader “Job Shock” thesis rather than a Khan Academy-internal trigger; it is included here as the contextual factor prompting the sensing/seizing response, not as an instance of a microfoundation itself.

The origin edition of the AWS “advanced team structures” franchise (Davis, AWS re:Invent, December 2025)

The wiki’s earliest-dated instance in the AWS Enterprise Strategy “advanced team structures” talk lineage that Allen’s London edition and Brovich’s Sydney edition already anchor on this page. Two elements are new to this earlier edition rather than carried forward: Stephen Brozovich’s three cross-cutting organizational tensions — speed, resourcing, connections — are a digital-seizing/strategic-agility instance (rapidly repositioning an organization along each spectrum as business context shifts), and his five-phase business-lifecycle-alignment model is a digital-transforming/redesigning-internal-structures instance (team shape — dedicated pods → hybrid → platform-plus-federated — evolving with customer maturity). Richard Davis’s Danske Bank case supplies a strategic-renewal/business-model instance (the “AI-first bank” ambition, 10 Big Wins) and a strategic-renewal/organizational-culture instance (the change-management/corporate-jargon-decoding program). See enterprise-ai-adoption for the fuller treatment of both.

Structured inquiry as digital-sensing at community altitude ( MIT Sloan CIO Symposium, May 2026)

Allan Tate’s keynote opening the 23rd annual MIT Sloan CIO Symposium is the wiki’s first instance of digital-sensing/digital-scenario-planning and digital-sensing/digital-mindset-crafting exercised at the altitude of a professional leadership community rather than a single firm. The symposium’s “inquiry framework” — frame the question via scenarios → examine competing interpretive lenses and actor tensions → translate insights into CIO-level decisions — is digital-scenario-planning formalized as a recurring, cross-firm community ritual, distinct from the internal-strategy-team version W&W’s original 27-interview sample describes. The symposium’s own pivot from one-time annual events to a “year-round knowledge ecosystem” (continuous learning model, collective intelligence as a capability) is digital-mindset-crafting exercised on an inter-firm cohort rather than a single organization’s workforce. The source also touches digital-transforming/redesigning-internal-structures (the CIO redefined as “designer of intelligence systems… architect of governance”) and strategic-renewal/organizational-culture (the shift from consuming insights to producing shared understanding) — see enterprise-ai-adoption for the fuller treatment.

Organizational capability as the differentiator, once access to AI is common ( MIT Sloan CIO Symposium podcast, July 2026)

Episode 16 of the same weekly podcast that produced the section above contributes four further cell instances, this time at the workforce-and-infrastructure layer rather than the community-ritual layer: digital-transforming/improving-digital-maturity (the Linux Foundation full-stack skills-gap finding — AI/ML engineering, cybersecurity, FinOps, platform engineering, cloud computing — read as the workforce-maturity gap organizations must close before AI investment pays off); digital-transforming/redesigning-internal-structures (Davenport’s “redesign work, management and decision making” prescription, quoted at length, plus the reshaping of developer roles toward architecture and systems integration); digital-seizing/rapid-prototyping (the “build AI-based pilots that actually work… learn what it takes to bring them into operations” prescription); and digital-sensing/digital-scenario-planning (the explicit 5-to-10-year horizon argument — organizational change, unlike technology change, cannot happen in three years). See enterprise-ai-adoption for the fuller treatment.

Collective sensemaking: the same community, three more cells ( MIT Sloan CIO Symposium podcast, Aug–Oct 2026)

The four-part Collective Sensemaking series from the same podcast describes the symposium’s own AI workflow and adds cell instances at a third layer, the community’s memory:

• digital-transforming/improving-digital-maturity: recorded conversations become indexed memory records a model can draw on (episode 20).
• digital-seizing/strategic-agility: the stated purpose of keeping dissent and decision history is to lower the cost of changing course, which Tate calls a hypothesis about “institutional self-correction” (episodes 20–22).
• digital-sensing/digital-scenario-planning: the Inquiry Framework in its full form, with question rewrites, three lenses and tiers (episode 21).
• strategic-renewal/organizational-culture and contextual/internal-barriers: Irving Wladawsky-Berger’s IBM example and his claim that scaling depends on an organization’s willingness to learn (episode 22).

All of it is a self-described prototype on the symposium’s own channel; no outcome data is given.

An independent second reach for the same organizational shape ( AWS NYC Executive Forum, July 2026)

Ishit Vachhrajani’s talk is the wiki’s first instance of the digital-transforming/redesigning-internal-structures and strategic-renewal/organizational-culture cells being exercised by the originating author of a framework the wiki previously held only secondhand — the four mental-model shifts Brozovich credited to him at re:Invent 2025 (silos → immune system; gates → guardrails; factory-floor → trading-floor; operational-execution → research-lab). The organization pillar’s accounts-payable worked example (a broadened-goal agent optimizing cash flow, not just task completion) and the internal-Amazon agent-org-design anecdote (role-boundary redesign after a 15-parallel-agents chaos phase) are redesigning-internal-structures instances independent of the AWS “advanced team structures” franchise; the barbell culture (top-down high-conviction investment + bottom-up reskilling) is a fresh organizational-culture instance. digital-seizing/rapid-prototyping (the AWS marketing AIR case, concept-to-GA in 90 days) and digital-sensing/digital-scenario-planning (the task-length-doubling / cost-of-intelligence-falling trend framing) round out the reading; contextual/external-triggers covers the Singapore Davos-2026 governance citation. See enterprise-ai-adoption for the fuller treatment.

A substrate-vendor instance of redesigning-internal-structures ( LangChain, July 2026)

Jensen Huang’s claim that “most companies will be built on harnesses” rather than business processes is the wiki’s first digital-transforming/redesigning-internal-structures instance argued from the substrate/silicon-vendor altitude rather than the enterprise-operator or consulting-firm altitude that dominates this cell elsewhere. The mechanism named is structural, not merely tooling-level: each proprietary workflow becomes its own autonomous harness, and the collection of harnesses is the company, replacing the collection of documented business processes. digital-seizing/rapid-prototyping is also present — the jointly-announced NVIDIA + LangChain Deep Agents + OpenShell blueprint packages model, harness, and runtime as a reusable starting point explicitly meant to compress the time from “domain-specific problem” to “deployed super agent,” in the same rapid-build spirit as the cell’s MVP/digital-innovation-lab operational definition, even though the blueprint itself is a vendor product rather than an internal lab. See enterprise-ai-adoption and agent-harness for the fuller treatment.

A first-party platform-vendor instance of redesigning-internal-structures and rapid-prototyping (Claude Platform team, July 2026)

The Claude Platform team’s own account of engineering-team reshaping — same headcount, most of the team now holding end-to-end design opinions while “orchestrating their Claudes” — is a first-party-vendor instance of digital-transforming/redesigning-internal-structures, complementing Huang’s substrate-vendor instance of the same cell: two different vendor altitudes (silicon/model vendor vs. first-party platform vendor) independently naming the same restructuring shape. digital-seizing/rapid-prototyping is also present, but in a distinct register from prior instances of this cell: Angela Jiang’s staged-ROI prescription (prove speed/productivity gains at individual scale before scaling to team, then company) is a lean-startup-style validate-small-first discipline applied to internal AI-adoption measurement itself, rather than to a product MVP — and the named hackathon system (“Urrea,” an agentic industrial-knowledge-capture system) is a conventional rapid-prototyping worked example in the industrial-ai-agents application class.

A streaming-media-incumbent instance of redesigning-internal-structures, improving-digital-maturity, and organizational-culture ( Netflix CPTO, July 2026)

Elizabeth Stone’s account of Netflix’s hiring and culture practices is the wiki’s first instance of these three cells at a large, public, decades-old streaming/entertainment incumbent, distinct from the banking (DBS, Danske), retail (DFI), industrial (Rolls-Royce), professional-services (PwC), and nonprofit-education (Khan Academy) instances already on this page:

• digital-transforming/redesigning-internal-structures — the hiring-mix shift toward systems thinkers and generalists across every function (not engineering alone), central/core-infrastructure engineering growing to build shared paved paths, and a universal AI-fluency career-ladder overlay applied identically from new hires through senior leadership, including hiring-process changes (AI-tool use now permitted in coding interviews).
• digital-transforming/improving-digital-maturity — investment in common infrastructure so individual teams don’t each rebuild source-of-truth-data access and guardrails from scratch, and AI-enabled distillation of decades of Netflix’s own institutional experiments and insights, previously gated behind tenured employees’ personal memory.
• strategic-renewal/organizational-culture — “excellence as an operating system”: talent density, high agency/autonomy, comfort with risk-taking, resistance to adding process after failures (blameless retros over checklists), and the keeper’s test — a culture doctrine Stone explicitly likens to how “the top AI labs operate,” though she frames it as a decades-old Netflix doctrine the AI era has made newly legible rather than a response built specifically for AI.
• digital-seizing/rapid-prototyping — PMs, designers, and data scientists getting further into the product-development life cycle (prototyping and testable code) before requiring an engineering handoff.

Notable methodological point for the concept: Stone explicitly frames Netflix’s culture doctrine as predating the AI era rather than as an AI-era response — closer in spirit to Erginbilgiç’s non-AI Rolls-Royce case than to the AI-native operationalisations elsewhere on this page. This strengthens the page’s working hypothesis (see Debates, below) that W&W’s cells are transformation primitives rather than AI-specific mechanisms — Netflix’s culture cell was already load-bearing before GenAI; the AI era changed what it needed to produce (systems thinkers, paved paths), not its existence.

An open-model-platform instance of sensing, seizing, ecosystem-navigation, and business-model renewal ( Hugging Face, July 2026)

Clem Delangue’s Hugging Face interview exercises five cells from a vantage new to this page — an open-source AI platform whose whole business is an ecosystem — complementing the substrate-vendor (Huang) and enterprise-operator instances above:

• digital-sensing/digital-scouting — scanning the open-vs-closed resurgence, the frontier-API cost curve, and China overtaking the US in open-model downloads (~41% per Hugging Face’s Spring 2026 report); screening frontier labs as digital competitors.
• digital-seizing/balancing-digital-portfolios — the core operating decision Delangue reports enterprises making: frontier APIs for experimentation/high-value tasks vs. owned/open models for production at scale (balancing internal and external options, setting an appropriate speed of execution — start off-the-shelf, then optimize/post-train).
• digital-transforming/navigating-innovation-ecosystems — Hugging Face is a digital ecosystem (the “GitHub for AI,” 16–17M builders, half the Fortune 500) that companies join and through which they interact with external partners; the platform strategy (“create 100× the value, capture 1–2%”) is ecosystem-navigation as a business.
• strategic-renewal/business-model — two business-model claims: the enterprise shift from renting to owning AI as a value-creation/value-capture logic change, and Hugging Face’s own contrarian, capital-efficient, community-first platform model (no funding round in 3 years, turned down an Nvidia investment).
• contextual/external-triggers — Chinese open models as disruptive competitors, the regulatory environment (US limits on private model releases), and concentration-of-power as the framing external forces.

The fuller treatment of the own-vs-rent / sovereignty / concentration-of-power theme lives on the new open-source-ai concept and in enterprise-ai-adoption.

Five video sources on the W&W cells, August 2026 — with balancing-digital-portfolios appearing twice in one week

The 2026-08-12 batch tagged five video sources against the W&W vocabulary, and two of them independently instantiate digital-seizing/balancing-digital-portfolios — the microfoundation the wiki has had the least concrete material on — in the same week and in strikingly similar form.

WorkLab describes a double-path allocation: the whole organisation gets upskilling and departmental experimentation, while a separate 60–120-person cross-disciplinary frontier unit receives a materially higher per-head token budget, tests every new model on release, works in teams of two to eight, and hands winners back to the wider organisation. The allocation rule is explicit — “you don’t really want 80,000 people spending thousands of dollars a day while you’re in the experimentation phase” — and it is paired with a second, McKinsey’s 1:5 ratio of $5 on people per $1 on tools, within which she names incentives as the underweighted line item.

McKinsey describe the same microfoundation as a sequencing decision: automate-first versus right-shore-first, “decided per process and even per part of a process — horses for courses,” against three named factors (existing capability and AI experience, impact expected and its timing, process type), with materiality thresholds set before a use-case list is drawn up. Both are portfolio-balancing in the W&W sense — allocating scarce budget and attention across bets of differing maturity — but one balances who experiments and the other balances what gets rebuilt first. Held together they are the wiki’s most operational material on this cell to date, and a useful complement to Bello’s corporate-venture-building evidence, which quantified the returns to portfolio breadth without describing the allocation mechanics.

The batch’s other cells: digital-transforming/redesigning-internal-structures appears in three of the five (Miller’s 34-agent org chart and near-flat frontier unit; the GBS diamond talent model replacing the pyramid; the BBC panel’s workflow-rethink-from-the-ground-up argument). contextual/internal-barriers appears in two, and in unusually concrete form — the BBC panel’s 3:1 tool-access-to-training ratio and “keys to the car without teaching them to drive,” and the GBS lighthouse that failed because the sales force would not adopt a technically excellent system. contextual/external-triggers appears in two from opposite directions: Collison’s incumbent buyers “springloaded to adapt” out of “a real terror of being left behind,” and Frey’s restrictions on foreign use of frontier models turning model dependence into a strategic exposure. And strategic-renewal/business-model appears in two — Miller’s cannibalize your business lines (with the analyst who built an energy-audit product priced at 50% of client savings) and the GBS three E’s reframing that moves shared services “away from just transactions and cost and more towards outcomes and value.”

A seven-source batch, and the first strong showing for digital-scouting and external-triggers

The second 2026-08-12 batch tagged five of seven video sources against the W&W vocabulary; two — Haghighi’s CS547 seminar and the CS329A lecture — were deliberately left untagged as design-research and LLM-internals material respectively, per the page’s when-not-to-tag guidance. Worth recording that the omissions are as deliberate as the tags: a corpus in which everything gets tagged would make the field useless as a filter.

contextual/external-triggers is the batch’s most-instantiated cell and, unusually, from three quite different directions. Ng: lobbying against open weights, data-centre moratoriums, and price-sensitive non-US markets defaulting to Chinese open models — external forces reshaping what American firms can build on. Hines-Pierce: a 6.5-million-unit global housing shortfall, an AI-driven energy supercycle, $1.5tn of funded projects stuck in permitting, and a skilled-trades shortage with no near-term supply response — external constraints that bind regardless of a firm’s own capability. And Frey: restrictions on foreign use of frontier models converting model dependence into strategic exposure. Held together these describe a period in which the binding constraints on digital strategy are increasingly outside the firm — political, physical and demographic rather than technological.

digital-sensing/digital-scouting gets its clearest instance yet from Frey, whose prescription for anyone behind the frontier is a scouting prescription with a quality criterion attached: adopt technology invented elsewhere (the Marshall-aid precedent), while recognising the binding constraints are institutional and cultural rather than availability-based — and that adoption for what decides whether it produces growth. “If people adopt it for email, it’s not going to drive growth in any meaningful way.”

The batch’s other cells: digital-transforming/redesigning-internal-structures in three sources — Brynjolfsson’s pyramid-to-diamond with its named consequence for the middle and senior ranks, Hines-Pierce’s middle governance team whose sole remit is making a solved workflow propagate, and Brown’s account of PwC cutting entry routes from “17, 18 ways you could join” to about five while hiring for potential over credentials. digital-transforming/improving-digital-maturity in Ng’s agent-ready-data argument (with a concrete failure mode: credential prompts breaking automated access) and Turnbaugh’s forty-year lineage of attempts to codify business meaning. strategic-renewal/business-model in Brynjolfsson’s find-new-value-rather-than-cut-cost argument and his small-business case. strategic-renewal/organizational-culture in Hines-Pierce’s 70%-people framing, three-generation stewardship doctrine, and explicitly Darwinian stance on who comes along. And contextual/internal-barriers in Brown’s seniorization gap (“it’s not going to happen by osmosis”) and Turnbaugh’s two barriers — standards that get supported but not adopted, and context that must be remembered to be invoked.

Sensing gets its most operational instance yet (Banholzer & LaBerge, August 2026)

The wiki’s digital-sensing/* material has generally been about posture — scanning widely, crafting a digital mindset. The competitive-advantage episode supplies the first source that specifies sensing as instrumentation, with a signal taxonomy, a cadence and a wiring diagram.

digital-scouting here means outside-in detection of seven advantage markers across the top 5,000 global companies, screened against three explicit tests (within-industry variation, actionable specificity, accuracy) — and honestly scoped as “an outside-in ten-minute scan.” digital-scenario-planning is the durability axis operationalised: batch-mode scans (patents, competitive offerings, rival capex) alongside always-on tracking (startup-activity spikes, regulatory news), with “triggers cascaded through the business” wired into two named destinations — the strategic scenarios, and whether resourcing decisions get revisited “maybe even not just once a quarter but on demand.” The cadence claim is the sharpest part: firms that do this well run it monthly to quarterly, not annually.

The CGM case gives the cell a worked example with an unusual lesson: the actionable signal was not the technology’s arrival but the regulatory and reimbursement changes that made the substitute purchasable, which “preceded the massive tip by a few years.”

balancing-digital-portfolios gains a hard allocation rule from the same source — the deficit asymmetry means fixing a deficit dominates pushing a strength further, since one deficit among the seven moves economic profit “from slightly positive to massively negative.” strategic-renewal/business-model gains the extensibility test (choose adjacencies where your advantage is valued, not merely applicable) and contextual/external-triggers the erosion thesis (an 11% shuffle-rate rise across 60%+ of industries; “the overall porosity of industry barriers”).

The Toyota clip adds two digital-transforming/* instances at target-firm altitude — improving digital maturity by making decades of dispersed research queryable, and redesigning internal structures via a central Enterprise AI function authoring per-function skills — though with no evidence of outcome attached.

A six-source batch that tests the digital- prefix from both ends (August 2026)

The 20 August 2026 batch tagged all six of its sources — the first batch in the corpus with no deliberate omissions — and it is unusually informative about the vocabulary’s reach because two of the six are non-AI or barely-AI sources.

The strongest non-digital test since Rolls-Royce ( GOTO Copenhagen 2025)

Where Erginbilgiç tested the non-digital reach of strategic-renewal/*, Rohrer tests it for digital-transforming/redesigning-internal-structures — and the result cuts the other way in an instructive manner. His entire talk is a structural-redesign argument (co-design organisation and system; restructure five layer-owning teams into one capability-owning team; the fractal team / team-of-teams model), and it exercises the cell completely, from a 2025 recording with essentially no AI content. The cell’s canonical activity list — “hiring a chief digital officer; digitalization of business models; designing team-based structures” — is satisfied only by its third clause.

That is evidence for the wiki’s working hypothesis in a sharper form than usual: the cell is really two cells wearing one label. Designing team-based structures is a transformation primitive with a literature stretching back through Conway (1968), Malan (2008) and Team Topologies; hiring a CDO and digitalizing business models are genuinely digital-era activities. Sources tagged against this cell should be read for which clause they exercise.

He also stretches digital-seizing/strategic-agility in a direction the cell description does not anticipate. “Pacing strategic responses” is usually read as speed; Rohrer’s punctuated gradualism reads it as a funding cadence — continuous ~20% allocation to complexity-debt paydown, with step changes treated as a reluctant exception (“second-system syndrome… most people don’t end up building a third system”) and the strangler fig as the mechanism for change that cannot be paused. And contextual/internal-barriers gains a barrier the cell’s wording misses entirely: complexity as accumulated entropy, asymmetric to remove (“if you put some complexity in, that can be fairly cheap. But reversing it out is asymmetrical”) and fatal if unfunded.

Wayfinding qualifies digital-mindset-crafting ([[2026-08-16-hill-bloomberg-leaders-ceo-skills-age-of-ai|Hill / Bloomberg Leaders]])

The most consequential single tag in the batch, because it constrains a cell rather than instantiating it. The cell’s first activity is “establishing a long-term digital vision”. Hill’s claim is that for breakthrough change the vision is not available — “many leaders have no vision about what’s really going to go on with AI. They just don’t. When you ask them, they’re perplexed” — and that demanding one is the wrong request. What substitutes is purpose plus wayfinding (“we don’t know the destination, nonetheless how we’re going to get there”), the declaration that “there is nothing called business as usual… everything is a working hypothesis”, and the beginner’s-eye staffing rule that operationalises the cell’s second activity (“enabling an entrepreneurial mindset”) as a procedure rather than an exhortation.

She also gives digital-seizing/strategic-agility its clearest mechanism in the corpus: decision rights, pre-agreed, justified by sensing latency (“somebody down there is going to hear it sooner than you”). And contextual/internal-barriers gains the unspoken board risk-appetite divergence — a barrier at the authorising layer rather than the operating one, which none of the cell’s canonical activities (rigid planning, hierarchy, change resistance) covers.

Two sources exercising redesigning-internal-structures from opposite premises

Blomfield argues hierarchy dissolves because AI removes the information-routing rationale (“no middle management”; ICs plus DRIs), and adds a distinctive improving-digital-maturity instance: legibility-as-house-rule (transcribe everything, ban Slack DMs, “every action needs to create an artifact… otherwise it basically didn’t happen to the AI”) with the office-hours-to-user-manual loop as “leveraging digital knowledge inside the firm” taken to its conclusion. His strategic-renewal/collaborative-approach instance is loop-to-loop coordination replacing human handoffs; his contextual/internal-barriers are hierarchy and committee (“I worked at a bank and we had so many committees for everything, and it grinds things to a halt”).

Rohrer reaches the same structural conclusion with no AI premise. Recording both against the same cell is the point: the convergence is not evidence for the AI mechanism, because the conclusion was reached and argued in detail before the AI-native-company literature reached it independently.

Portfolio balance gets a per-capability decision grid ( Sequoia)

digital-seizing/balancing-digital-portfolios — “balancing internal and external options” — gets its most literal instance yet: own-vs-rent decided per capability against cost, latency, in-domain performance and data proprietariness, with a worked case where the two answers land on opposite sides inside the same product (rent the coding agent, own the tab-autocomplete). Her digital-scenario-planning instance is the centralized-vs-decentralized-intelligence framing; her redesigning-internal-structures instance comes with a named anti-pattern (“a lot of companies are shoehorning this AI platform team into doing the sovereign AI stuff as well. I’d encourage folks not to do this”); and strategic-renewal/business-model is “the product is the intelligence”.

Two narrow tags, deliberately

Anthropic takes only rapid-prototyping and improving-digital-maturity; Google Cloud takes one cell only (improving-digital-maturity, for vocabulary-levelling among builders). Both could have been stretched to more; neither was. Consistent with the note above about the batch’s deliberate omissions, restraint on genuinely engineering-flavoured sources is what keeps the field usable as a filter.

Debates and supersession

• Bounded programmes vs. pure adaptability (added 2026-09-21). Ross & Schneider argue that transformation and change management presuppose a fixed destination and should give way to adaptability. Ellmer and Dhar agree the destination is unknown and adaptability is the capability, but say bounded chapters with visible endings are how people sustain continuous change. Both are practitioner claims; neither cites an outcome study.

• Does the DC→performance link hold for economic performance? Cimino et al. 2025 is the corpus’s first quantitative test, and its one clean failure is H2: dynamic capabilities do not significantly predict economic performance in 210 Italian innovative startups (β = 0.138, CI [−0.013, 0.281]). Every other DC path in the model holds strongly. The authors read it as a stage effect — capabilities in early-stage firms are aimed at experimentation, product development and market entry, so financial return lags capability development — and align it with Wilden et al. (2013), who found the link conditional on organic structures and competitive intensity. No supersession. This is a boundary condition on the concept, not a refutation of it: DC→technological (0.447), DC→collaborative (0.615) and DC→sustainable (0.662) performance all hold in the same model. What it rules out is the loose claim that dynamic capabilities pay off financially as such, independent of firm stage.

• Sensing-as-prediction vs sensing-as-shock-readiness. The Teece (2007) framing of sensing emphasises opportunity and threat detection — close to forecasting language. Erginbilgiç 2026 argues against the prediction-framing: “It’s not about actually predicting the world, it is about how your company now thinks about dealing with external shocks” (~21:14–21:32). Warner & Wäger 2019’s digital-scenario-planning microfoundation is closer to Erginbilgiç’s shock-readiness framing than to pure forecasting. No supersession; the productive tension is between sensing as accuracy and sensing as response capability. The wiki currently treats them as compatible (sensing must produce both signal-detection and the organisational habit of responding to signals).

• Digital vs non-digital scope of W&W cells. Warner & Wäger 2019 derives its cell vocabulary from digital transformation case studies; Erginbilgiç 2026 is a pure non-digital case that nonetheless maps gracefully onto strategic-renewal/organizational-culture, digital-seizing/strategic-agility, and contextual/internal-barriers. The cells stretch outside the digital lens with the digital-mindset clause optional. No supersession; the wiki’s working hypothesis is that W&W cells are transformation primitives whose digital-flavour reflects the empirical setting of the original 27-interview sample rather than a load-bearing scope restriction.

• Vendor-altitude vs operator-altitude vs case-study altitude. AWS supplies vendor-altitude operationalisation; DFI supplies operator-altitude case material; Erginbilgiç 2026 supplies CEO-altitude case material at a non-AI incumbent. No contradiction across the three; the wiki triangulates the dynamic-capabilities concept across all three altitudes deliberately.

Related concepts

• enterprise-ai-adoption — AI adoption is a contemporary instantiation of digital transformation; the same sensing/seizing/transforming logic applies, with AI-specific subcapabilities.
• generative-ai — extends the digital-transformation context; GenAI is a current sensing/seizing target for incumbents.
• automation-vs-augmentation — a strategic-deployment choice that lives within seizing capabilities.
• strategic-foresight — sensing-cluster microfoundation; FTSG-style methods are sensing tools.
• systems-thinking — adjacent lens for transforming-cluster decisions about flows and ecosystem boundaries.
• MIT CISR Four Stages — staged-maturity view of digital/AI transformation; complementary to the dynamic-capabilities lens.
• Tin Man — adjacent framing of org design under environmental change.

Open questions

• The Warner & Wäger study is from 2019, pre-GenAI. How do the nine microfoundations need to be updated for the 2026 GenAI context? (Open question; possible synthesis topic.)
• Cross-source mapping: MIT CISR’s Four S (Strategy/Systems/Synchronization/Stewardship) and the dynamic-capabilities framework appear to overlap substantially in scope but use different vocabularies. Would benefit from a future synthesis page.

A seizing cell, measured: can a model balance a digital portfolio? (added 2026-09-09)

[[2026-03-11-allen-mcdonald-how-well-can-ai-do-strategy-simulation-benchmark|Allen & McDonald (Strategy Science, 2026)]] is, as far as the corpus goes, the most direct empirical measurement of a single Warner & Wäger cell — digital-seizing/balancing-digital-portfolios.

The Back Bay Battery simulation is that cell rendered scorable: a fixed R&D budget, a profitable but declining core technology, an emerging one that only pays off after early and sustained commitment, and no instruction about what to maximise. Getting the balance right is the whole task, and the failure modes map onto the cell’s own logic — “spreading investments too thin leads to mediocrity, ignoring disruption leads to decline, and over-committing too early leads to collapse.”

The finding is that the newest models fail this cell in one specific direction. Mid-to-late-2025 frontier models (GPT-5, o3, Gemini 2.5 Pro) achieve high cumulative profit with very low emerging-technology revenue — they optimise the core and decline the portfolio balance — scoring below both earlier models and MBA students. Late-2024/early-2025 models did not fail this way; they exceeded the human average by timing investment and pricing to get both.

Two things follow for this page. First, the microfoundations argument gets empirical traction: the authors conclude that “the microfoundations of successful strategic decision making may differ in important ways from those underlying success in other domains like coding or advanced science” — which is a claim about which capability is being measured, and a warning against treating general capability as a proxy for a specific one. Second, and more practically, it is evidence about where an AI-assisted seizing process needs a human: not in the analysis, which these models do competently, but in the commitment to an option whose payoff is not yet visible in the data. See the source page for the limits, including that the study measures reasoning over supplied information rather than agentic sensing.

The spine, measured: the first quantitative test in the corpus (added 2026-09-17)

Until now this page rested entirely on conceptual work (Teece 1997, 2007, 2014) and on interpretive case research (Warner & Wäger 2019’s seven German incumbents and 18 consultant interviews). Cimino, Corvello, Troise, Thomas & Tani (2025) is the first source in the wiki to contain a path coefficient at all — PLS-SEM on 210 Italian innovative startups, with sensing, seizing and transforming modelled as a reflective second-order construct.

The measurement result matters independently of the paper’s own thesis. The three-cluster construct loads cleanly (Cronbach’s α 0.832, rhoC 0.900, AVE 0.750), which is evidence that sense/seize/transform is a coherent latent factor and not an artefact of how qualitative researchers code their interviews. That is external support for the spine this page is built on — though note it is support for the generic Teece three clusters, since the instrument (Kump & Kessler 2019) contains nothing digital-specific. It says nothing about W&W’s nine digital microfoundations, which remain qualitatively grounded.

The substantive finding is a readiness–implementation asymmetry. The paper splits AI adoption into acceptance (AIAR — openness and readiness) and intensity (AIAI — how far AI is actually embedded in processes), and finds dynamic capabilities produce acceptance nearly twice as readily as implementation: DC→AIAR β = 0.523 against DC→AIAI β = 0.279. Only the implementation leg reaches economic performance. For this page the implication is that dynamic capabilities are better understood as producing the conditions for adoption than adoption itself — the capability generates willingness; converting willingness into embedded practice is a separate, harder, resource-bound step. Two caveats travel with this: the two adoption constructs correlate at HTMT 0.878, uncomfortably close to collinear, and the sample is 81% micro-firms under eleven employees. See the source page for the full reliability read.

The rival firm-level lens: TOE (added 2026-09-18)

The adoption literature’s firm-level theory is not dynamic capabilities but the technology–organization–environment framework (see technology-adoption-theories). Schwaeke et al. (2024), mapping 106 studies of SME AI adoption, make the one explicit link between the two: their culture cluster credits leadership with “a strong role in the dynamic capabilities of firms through the ongoing development of new practices.”

The two answer different questions and should not be merged. TOE explains whether a firm takes up a technology, from its technological, organisational and environmental context. Dynamic capabilities explain whether a firm can renew itself. Several TOE clusters read naturally as W&W cells — the ecosystem cluster as navigating-innovation-ecosystems, the knowledge cluster as improving-digital-maturity — but the mapping is the wiki’s, not the review’s. Nguyen et al. (2026), ingested 2026-09-21, fit both into one model. TOE factors (top leadership and government support) drive the level of AI adoption, and dynamic capabilities are invoked to explain how that adoption turns into business-model innovation, competitive advantage and performance. It is the mirror of Cimino et al.: capabilities follow adoption rather than precede it. But the capabilities are theorised, not measured, and the outcome paths explain little (R² 0.04 for business-model innovation, 0.06 for competitive advantage). Its one distinctive result is that deeper AI adoption strengthens the link from an existing advantage to performance — AI as amplifier, not source.

Concentration needs a capacity to pivot; continuous change needs endings (added 2026-09-21)

Two practitioner sources describe the capability side of AI change.

• IBM on Bain’s Winning with AI, September 2026: IBM’s CEO argues for concentrating on a few AI bets (seizing). Bain’s Sarah Elk adds the condition: concentrated bets can be wrong, so they only work in firms that test and change course — “acknowledging a mistake in 6 months is far better than going 3 years and getting a mediocre result.” This is digital-seizing/strategic-agility: portfolio concentration and the capacity to pivot as one capability rather than two. IBM’s own story is the negative case. After Watson it built monolithic vertical applications in health, a domain where it knew neither customer nor regulator.
• BCG, August 2026: “Adaptability is something that we can learn, and you can build it into the system.” But they argue that continuous change still needs constructed endings, because “humans need endings” and fatigue lowers the capacity for the next change. That is a claim about how the transforming capability is renewed: bounded chapters inside continuous change.

Mobilization as a named part of seizing (added 2026-09-21)

Teece’s table above defines seizing as “mobilizing resources to capture opportunities”. McKinsey, September 2026 take that verb and make it the whole argument. Their claim is that mobilization is distinct work between strategy and execution: owned initiatives, reallocation of talent, capital and agentic budget, and commitments carried into budgets. They also claim it is what most distinguishes firms that rise on McKinsey’s power curve from firms that fall.

Three points connect to this page:

• digital-seizing/balancing-digital-portfolios, with a magnitude attached. Climbers invest about twice what their average competitor does in their chosen priorities. The prescribed method is to stack-rank bold moves against everything else being funded, rather than judge them one at a time.
• contextual/internal-barriers, named as cognitive biases. Anchoring to last year’s budget and loss aversion in single-move evaluation are the two barriers named. Each comes with a procedural fix, so the barrier is treated as something a process can remove rather than as culture.
• digital-seizing/strategic-agility, as a reinforcing loop. Firms good at test-learn-adapt are more likely to be good at mobilization, “because you have to mobilize to learn”, and firms good at mobilization are more likely to hold strategy to a high standard. At its most mature, mobilization is routine: “everyone knows what happens next.”

This is practitioner research from one firm, and the underlying studies are not ingested. It supports the seizing microfoundation without testing it, so confidence is unchanged.

Sensing, rewiring, lock-in: the triad renamed for AI (added 2026-10-15)

Lakhani, Stave, Ng & Martines (HBR, July 2025; read here through the HBS AI Institute’s summary) call the missing ingredient in AI value change resilience: “the ability of companies to adapt, reinvent, and scale new ways of working.” They split it into three “muscles”, which map one-to-one onto Teece’s triad and Warner & Wäger’s digital version:

Change resilience	Definition (as quoted)	Dynamic-capabilities counterpart
Sensing	”pick up weak technological, competitive, or societal signals early”	sensing; W&W digital-sensing/digital-scouting
Rewiring	”redeploy talent, data, capital, and decision rights in days or weeks, not fiscal quarters”	seizing; W&W digital-seizing/strategic-agility
Lock-in	”codify what a team learns (in process, code, or policy) so the next initiative starts from a higher baseline”	transforming; W&W digital-transforming/redesigning-internal-structures

The summary cites neither Teece nor W&W. Two things are added. The first is a time scale, days or weeks rather than quarters, which W&W leave unstated. The second is lock-in as codification, which reads the transforming cell as organisational learning that compounds rather than as a one-off restructuring. The five-step playbook (Learn, Do, Imagine, Act, Care) adds Care, employee wellbeing during continuous change, which no W&W cell covers directly. A practitioner framework from HBS’s AI Institute arriving independently at the same three-part shape is further evidence that this page’s spine is the natural decomposition. It is not new evidence that the capabilities produce AI value: the only number in the summary is BCG’s figure that 26% of organisations have achieved value from AI.