Legacy Software vs. AI - Superposition Fallacy
The market sees through the fallacy that legacy enterprises can decouple from legacy software.
AI-Native Means Recapitulate Entire Company
Grafting AI to a legacy enterprise proves no different than this:
These ships proved highly dispensable, once purpose-built coal steamers came into the fore. Even in the 19th century, this design did not last even 50 years.
Likewise, fusing agentic AI into legacy enterprises, with their diffused and often conflicted boards and politically-appointed executives, completely defeats the purpose and potential of agentic AI.
AI-Native Means Devolving Decision-Making to the AI, Not Suits
Automating labor produces marginal returns and actually shifts inordinate amounts of validation work to senior managers.
Instead, improving strategic decision-making produces outsized returns, without flooding manpower with validation work and creating "trust" bottlenecks.
SNARKs Matter: AI-Native Leadership Looks a Lot Like Trustless Computing
Remember all the crypto hype about "trustless computing"?
Now, we finally need it: for validating work done by AI agents in the service of the Enterprise.
Why? When agents LITERALLY DECIDE THE FATE OF THE COMPANY, only a trustless validation can prove that the agent did the work with integrity - no injections, steering, et cetera occurred during the agent's run.
Superposition Fallacy: Legacy Software Didn't "Recover" - It Moved Back in-Line with Legacy Enterprises
The ONLY thing that corrected, recently, with software shares rebounding, was the spread between software market capitalization and that of their legacy enterprise customers.
BOTH face fundamental threat from agentic AI and both should react to agentic improvements and developments, rather than only the software component, in isolation.
They are not independent: there is no superposition of legacy software with legacy enterprise - they are both legacy, entirely, with respect to agentic AI.
Which Enterprises Are Truly AI-Native?
An enterprise which devolves strategic decision making to agentic AI is one that has become AI-native.
The rest are hybrid steamships.