Science & Tech · AI · 5 hrs ago
Enterprise AI still lacks a breakthrough architecture
The Transformer architecture underpins GPT and much of generative AI after it was defined in a 2017 paper by eight Google researchers.
Google used the technology in products such as Search and Translate, while OpenAI built GPT and later ChatGPT on it.
The author argues that enterprise AI needs a breakthrough beyond more capable individual AI tools.
A person or AI can complete tasks, but a company also needs to coordinate work, keep information over time and adapt.
In a 2026 McKinsey survey, 80% of respondents said AI improved their personal productivity, but 37% reported a positive effect on company earnings.
McKinsey found that companies that redesign workflows tend to get better results than those that simply add AI to existing processes.
The author says current systems focus on connecting AI agents, but it remains unclear how to make them work as an intelligent organization.
Enterprise AI has yet to show a breakthrough architecture that turns capable AI models into intelligent organizations, according to a Fast Company article.
The article says the Transformer architecture, introduced in a 2017 research paper by eight Google researchers, underpins GPT and much of the generative-AI wave.
It argues that individual AI task performance does not by itself give companies the ability to preserve state, coordinate capabilities, enforce authority, remember outcomes and adapt over time.
A 2026 McKinsey survey cited in the article found that 80% of respondents reported improved personal productivity, while 37% reported a positive contribution to enterprise EBIT.
The article says companies that redesign workflows are typically among those getting stronger results, and describes current agent-based approaches as useful but insufficient to design an intelligent organization.
- Who
- Enterprise organizations seeking to use AI, and the companies developing AI systems and tools.
- What
- The article argues that enterprise AI still lacks an architectural breakthrough to turn AI capabilities into organizational performance.
- When
- Published November 8, 2026. It discusses the Transformer breakthrough in 2017 and findings from a 2026 McKinsey survey.
- Where
- Not stated
- Why
- The article says task-capable AI agents do not by themselves provide the coordination, memory, authority and adaptation needed for organizational performance.
This story does not have two clearly opposing sides.
OpenAI was founded two years before the Transformer paper appeared.
Eight Google researchers wrote the paper “Attention Is All You Need,” which defined the Transformer architecture.
A McKinsey survey found that 80% of respondents reported improved personal productivity and 37% reported a positive contribution to enterprise EBIT.
Fast Company published the article arguing that enterprise AI is still waiting for a Transformer-scale architectural breakthrough.
- Transformer paper
- Attention Is All You Need,” published in 2017
- Paper authors
- Eight Google researchers
- Personal productivity
- 80% of respondents in McKinsey’s 2026 survey
- Positive enterprise EBIT contribution
- 37% of respondents in McKinsey’s 2026 survey
- Article publication
- Fast Company, November 8, 2026
What people are saying
Is coordinating agents the same thing as designing an intelligent organization?
To me, at least, it looks completely different.