James Gross, July 9, 2025
In his farewell address on January 17, 1961, President Dwight D. Eisenhower introduced the term "military-industrial complex" to caution against the growing influence of a powerful alliance between the U.S. military and defense contractors. He warned that this partnership, while it would contribute to national defense, posed a potential threat to democratic governance if left unchecked. Eisenhower emphasized the need for an "alert and knowledgeable citizenry" to ensure that this complex did not acquire unwarranted influence over American policy and society.
Just six years later, the programmer Melvin E. Conway observed in 1967 that “any organization that designs a system … will produce a design whose structure is a copy of the organization’s communication structure.” In plainer words: the way people and teams talk to one another inevitably shapes the architecture of the software, product, or process they build.
Conway introduced the idea in his Datamation article “How Do Committees Invent?” (1968). He noticed that when a compiler was written by one small group, it was a simple one-pass design, but when responsibility was split between two groups, the compiler became two-pass, mirroring the organizational split. Fred Brooks later popularized the adage as “Conway’s Law” in The Mythical Man-Month and has since been validated repeatedly in software projects and beyond.
Fast‑forward to the current SaaS era we are living through. Thousands of specialized SaaS platforms now form a “software-industrial complex.” Much like the military-industrial complex reshaped U.S. defense, this ecosystem is quietly rewriting corporate structure and often creating more problems than it solves. Pre-configured approval chains, segregation-of-duty rules, SLAs, and escalation paths embed governance assumptions into everyday workflows, nudging companies to adopt them wholesale. Once a vendor’s schema becomes the canonical record, its data gravity pulls every upstream and downstream team into orbit—or strands them behind translation layers that soon ossify. Over time, practitioners start speaking the platform’s language, deepening cultural drift and lock-in and forcing firms to hire talent fluent in an ever-narrower SaaS vernacular.
If the SaaS Industrial Complex molds organizations in its image, AI gives enterprises a chance to turn the mirror back to them: to design software that truly reflects their unique shape, ambition, and edge.
Unlike static SaaS platforms, AI is not a fixed product; it is a capability—a malleable, context-aware intelligence that can learn your language, follow your logic, and adapt to your expertise. We refer to this expertise as “private tokens”: proprietary organizational data and institutional knowledge that cannot be replicated by competitors or accessed by generic AI systems. AI now empowers enterprises to build systems that match their DNA, rather than contorting themselves to fit a vendor’s generic blueprint.
When you own the expertise, the training data, and the surrounding logic, your software becomes proprietary again—a strategic asset, not a shared commodity. Your private tokens stay private. Your workflows stop drifting toward the lowest common denominator.
In a world where every company has access to the same SaaS tools, the only way to stand out is to build something only you can build. That is, software that knows your org chart, your market, your language, and your people better than any off‑the‑shelf alternative ever could.
In learning from Eisenhower and Conway, before you sign the next multi-year SaaS contract, ask yourself: Whose communication structure will my company inherit—and who really profits from that design?
No matter what decision you end up making, ensure the answer to the question is unequivocally yours.