Post by **Ethan Mollick** published on **October 1, 2026** on *One Useful Thing* (~2,000 words), subtitled *Benefitting from the Bitter Lesson*. Mollick revisits an anticipation he now considers mistaken: he thought that getting agents to work in groups would require lengthy human design, "akin to building a company."
#Bitter Lesson#bitter lesson#agent management
Ethan Mollick — professeur à la Wharton School (University of Pennsylvania) · auteur de *One Useful Thing* (Substack).
SFEIR analysis note that reexamines the software architect profession in the age of generative AI through the framework of **Gregor Hohpe** (*The Software Architect Elevator*). Central thesis: the « **Oracle** » architect — the holder of supreme knowledge dictating rules from an ivory tower — is obsolete, since AI generates code and proposals on demand; the modern architect becomes an **intelligence amplifier (IQ Amplifier)** who provides teams with mental models, business context, and decision tools to leverage AI while ensuring system coherence. The document breaks down the impact **floor by floor of the "Architect Elevator"** (Enterprise / Solution / Platform / Software architect) and argues for **Domain-Driven Design (DDD)** as an essential safeguard: the **ubiquitous language** serves as the basis for *system prompts* (a domain dictionary injected via `.clinerules`/templates, reducing hallucinations and business misinterpretations) and **bounded contexts** restrict the scope entrusted to AI to maximize generation reliability. Conclusion: AI is not a threat but a catalyst that relieves the architect of technical grunt work to emphasize synthesis, strategic vision, modeling, and the human link between tech and business. Domain: software architecture, the architect's role, DDD, structured prompting, enterprise AI governance.
#Software architect#architect's role#generative AI