iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Generative AI is advancing quickly, but the systems and institutions that let organizations build durable businesses around it are still unsettled. That distinction matters: a capable model is not, by itself, a stable platform. Companies should learn quickly, keep early commitments reversible, and invest in capabilities that remain useful as the AI architecture changes.
What does an unfinished AI foundation mean?
In Kevin J. Boudreau’s August 26, 2026, article for MIT Sloan Management Review, “platforming” means creating the structures that allow organizations to build confidently on a technology. Generative AI may be progressing rapidly while that broader process remains incomplete. The publisher describes the tension as a fast-moving technology whose surrounding architecture for economywide transformation has not yet settled. MIT Sloan Management Review
The distinction is between model progress and ecosystem maturity. A model can become more capable without settling how applications should be built, how organizations should deploy them, or which parts of the system will become durable foundations.
Which parts of the AI stack are still changing?
The emerging stack includes specialized hardware, cloud computing, foundation models, and applications. The lower layers are more recognizable; the application and deployment layers remain fluid. Organizations are experimenting with model APIs, chatbot interfaces, agents, middleware, AI embedded in existing products, and enterprise deployments. These approaches are experiments, not proof that every interface or business role will endure. O’Reilly’s listing and preview of the article
#1 Best Overall
As a result, choosing an AI tool is not merely a model comparison. It is also a choice about interfaces, integration, provider reliance, and how much of a workflow the organization is willing to restructure around an architecture that may change.
Why does platform uncertainty complicate investment?
Boudreau’s analysis identifies several connected risks. Returns on complementary products are uncertain when the underlying architecture is unsettled. Switching costs may be limited, and customers may use more than one model. Training and inference have costs; open-weight models add competitive pressure; and shared models can make imitation easier as well as innovation.
These are strategic considerations, not universal economics: their relevance depends on the organization, its use case, and the alternatives available. The competitive tension is captured in the sentence attributed to Boudreau in a preview of the article: “The same foundation models that reduce the cost of innovation also reduce the cost of imitation.” O’Reilly article preview
Free tools Windows power users keep installed
One-click scans. No signup required.
How should a company invest before the platform settles?
Separate model capability from organizational value
Evaluate what a model can do, then separately ask whether the workflow creates value in your organization. Integration, processes, skills, and institutional context affect whether capability becomes useful. A model comparison alone cannot answer that question.
Keep commitments reversible where architecture is fluid
Prefer trials and designs that allow a team to change models or interfaces without rebuilding the entire workflow. The article’s publisher summary recommends learning faster than committing and building assets that can survive architectural change. That does not mean avoiding investment; it means matching the scale and permanence of a commitment to the maturity of the layer being adopted.
Invest in complements that travel across models
Look for organizational capabilities and products that retain value if the underlying model changes. Access to a widely available model capability may not, on its own, create a lasting differentiator. Consider whether the advantage lies in domain knowledge, workflow integration, data practices, or another complement that is genuinely valuable across alternatives.
Test where differentiation and dependence come from
For each proposed use, ask whether it depends on one provider, whether switching is practical, and what makes the result distinct from what competitors can build using shared models. The answers help distinguish a durable capability from a feature that may be easy to replicate.
A practical framework for comparing AI investments
Use these questions to assess options. They are decision axes synthesized from the article’s argument, not a ranking supplied by Boudreau.
Best Value
| Decision axis | Question to ask | What a stronger answer indicates |
|---|---|---|
| Durability | Would this capability remain useful if the model, interface, or deployment pattern changed? | Value is not tied entirely to one unsettled layer. |
| Provider dependence | Can the organization change providers without disrupting the whole workflow? | There is a practical path to adapt as alternatives evolve. |
| Differentiation | What value comes from the organization’s complements rather than shared model access? | The investment has a reason to matter beyond widely available capability. |
| Switching costs | What would it take to migrate the workflow, and are those costs justified? | The team understands both lock-in exposure and the cost of flexibility. |
| Integration | Can the organization incorporate the technology into its processes and responsibilities? | The investment addresses how work is done, not just what a model can produce. |
What the argument does—and does not—establish
The article presents a strategic interpretation of a changing technology ecosystem; it is not an independent evaluation of AI systems or proof of outcomes for every organization. A preview mentions an estimate of about 2.4 billion monthly users of generative AI platforms, attributed only to “some estimates.” Because it does not identify the original publisher or report year, that figure should not be treated as a fully sourced statistic.
The core practical implication is narrower and more useful: do not confuse rapid progress in models with a settled platform. Invest in learning, complements, and integration while preserving room to adapt where interfaces and roles remain uncertain.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Recommended Free Tools
