In the name of the Anode, the Cathode, and the Holy Jensen
The Molecule & The Grin
It was in that precise moment that I knew we had reached the peak AI hype cycle.
I was watching Nvidia’s CEO, Jensen Huang, interview on the Dwarkesh Patel podcast, when it was interrupted by a YouTube ad. Tony Robbins’ gleaming white grin appeared, and he bombarded me with hyperbolic claims about AI in an effort to peddle tickets and memberships to his latest event and course schemes.
With urgency, Tony warned me that we’re experiencing “more change in humanity than we’ve seen in all of our lifetimes.”
Panicked and horrified by Tony’s slick pitch, I desperately reached for the “Skip Ad” button.
Returning to the podcast, Jensen almost seemed to pick up where Tony had interrupted, but, where Tony used fear and sensationalism, Jensen offered eerily theocratic, biological and evolutionary claims.
For Jensen, he wasn’t just selling computer hardware, he was selling the next evolution of humanity:
“Making that token is like making one molecule more valuable than another molecule, making one token more valuable than another.”
When we start treating computing systems as “valuable molecules”, we stop treating AI for what it really is: a tool. We stop asking whether or not the output is truly valuable, or what the workflows and teams should look like to operate it.
The Super Tool of the 21st Century (again)
Jensen fetishes the token as “the transformation from electrons to tokens is such an incredible journey”. But, this journey has been around since I was in diapers (only by other names). Data processing, and in a marketing context, database marketing was another “transformation of the electron” that brought its own share of outlandish statements. For example, Oracle’s founder had this to say about the relational database:
“In 1977, everyone said I was nuts when I said we were going to build the first commercial relational database… If you defend a really big idea that challenges widely held beliefs, you’re likely to generate a mass of hatred… When Galileo defended Copernicus, he was ridiculed, imprisoned, and then threatened with death unless he recanted.” — Larry Ellison (Reflecting on the 1977–1980 era)
The supposed Copernican revolution of the relational database was touted by its inventor, Edgar F. Codd as the “Final Architecture” of information. Similar to Huang, Ellison used the language of natural selection to describe the change: “it is not sufficient that I succeed — everyone else must fail.”
By 1992, the “incredible journey” of the “Final Architecture” of the relational database had matured inside corporate enterprises like Bell Canada (a major telecommunications firm), where my late father led a database marketing team. While he was in this role, he handed me a copy of the below book on the subject with the convincing subtitle of “the Super Marketing Tool of the 21st Century”:
The relational database unlocked new opportunities for more detailed segmentation, which would sound familiar to anyone working with contemporary personalization, “customer 360” and “customer data platform” solutions. While large language models were decades away, many descriptive and predictive models (like customer lifetime value) were widely used. These are capabilities that are often marketed as “AI” today — they are well-known problems, and that’s why the supporting processes and roles of the old tools are genuinely applicable for the new.
Artistry (with people, not GenAI)
Jensen’s digital deism proselytizes the “artistry” involved when the electrons pump through his GPUs:
“The amount of artistry, engineering, science, and invention that goes into making that token valuable, obviously we’re watching it happen in real time.”
Today, artistry is required, but not in the way Jensen describes. The artistry we need is the same kind that my Dad advocated for with his team: workflows, teams, roles, and ownership. We need an AI exorcism where we can lose the cult and talk about what actually needs to be in place to adopt the latest tool within the enterprise.
With a little artistic flair, we can take inspiration from the workflows and roles of the past, while seeking out new human-centric ways of harnessing the latest AI technologies.
KNIME Exorcism
At the moment, the most hyped claims about AI are focused around “agentic” use cases, which has almost metastasized into its own Jensen-Robbin-like lexicon. Meaningful adoption of this technology can’t be a developer-only activity; it needs to be wrapped in processes, workflows and roles that helped tame the “Super Tools” of the past.
KNIME is a low code visual solution, which helps to bring multiple actors together to truly operationalize agentic AI. Developers, data scientists and business analysts are all first class citizens. Together, they can collaborate to adopt the processes that make agentic possible within the enterprise:
- “Human in the loop” circuit breakers and validation
- Document agentic logic with “explainable by design” workflows
- Auditability at every step (not just inputs and outputs)
- Maintain data sovereignty & security
- Create agentic non-linear flows
- Separating models from tools, data and other guardrails
- Environment isolation and version control
- Explainability for regulatory compliance & ethics
- Processes for sharing “advanced” code with business users
Jensen Huang argues that the “evolution of humanity” is the transformation of electrons into tokens. But as the history of the relational database shows us, the real evolution happens when we stop treating a technology as a miracle and start treating it as a responsibility.
We don’t need a “final architecture” or a digital deity. We need the same thing my father needed in 1992: a way for people to work together to ensure that the tools serve the enterprise. By focusing on what humans-in-the-enterprise need, we can finally complete the exorcism-leaving the hype behind and keeping the utility of the new tool.
Explore KNIME
To learn more about KNIME’s approach to agentic AI, visit “What is Agentic AI?”. If you are looking for free courses and the opportunity to build hands on, check out “Data Aware Agentic AI — Getting Started Course”. Follow and connect with me to stay up to date on what I’m building with KNIME.