Pursuant to Cromulent Marketing Incorporated’s moral obligations and general operating principles, we hereby declare our corporate policy regarding the use of generative AI (whether or not based on Large Language Models, more commonly referred to as “LLMs”):
We don’t use it. At all. For anything.
Wait, why not?
Goodness, you’re putting us on the spot. But off the top of our heads, here are some reasons:
- We want to continue to develop our domain knowledge, across any and all domains we encounter. Not only does our continued success as a marketing agency (consultancy?) likely rely on doing so, but — frankly — we like learning.
- We want to spend our time thinking and creating, not verifying that research is sound (e.g., comprehensive, unbiased, current, correct, etc.) or validating generated output (e.g., for validity, to guard against plagiarism, etc.)
- Our work generally pertains to cutting-edge and proprietary information, which limits AI’s utility and amplifies its shortcomings (e.g., hallucinating/MSU, uncredited verbatim reproduction, etc.). For example, our output is often the first time a particular subject has been captured and published in a written form, so specifics wouldn’t even be available in training data or via RAG.
- We find tremendous value in working through data the old-fashioned way. Could we use AI to crunch data (e.g., survey results, operational telemetry, etc.)? Sure, but crunching data is how we understand that (those, for the pedants!) data and the story being told, leading to “Aha!” moments and interesting paths of inquiry. It’s also how we can uncover errors in the datasets (whether mathematical or generally not passing the experience-informed sniff test). Plus, spreadsheets are fun (sorry, we’ve never had to use R or MATLAB in a professional capacity).
- Personal. Moral or ethical? Principled? Or maybe it’s simply that we don’t want to.
Ummmm…
Worry not!
Over the years, we’ve written about AI too many times to remember, covering probably all its guises and many closely related subjects:
- Machine learning? Love it
- Neural nets? Puh-leez, built those back in university, just as the AI winter was thawing
- Fuzzy logic? We got cozy
- Different learning/training approaches? We always thought adversarial techniques were especially cool, but we’re not against back propagation
- As presenting a potentially existential threat to humanity? Actually no, but we expect that to change pretty soon.
What are you doing with that AI? Here’s a partial list of fun applications we’ve written about:
- Simulating wind effects in built/urban environments, as an alternative or complement (usually a precursor) to computational fluid dynamics (CFDs) and wind tunnel testing
- Predicting electricity prices, enabling companies to optimize their utilities expenditures by shifting loads to take advantage of lower-priced periods and government incentives
- Modeling epidemics and pandemics
- Explainable AI, including how to peer into that black box, the value of that understanding, and how to productize it
- Cybersecurity? We literally struggle to recall all the different use cases … but the most recent two (as of October 2026) are an agentic swarm that performs detection and response in automated security operations centers (SOCs), and a mechanism to detect, respond to, and prevent malicious dependency-induced supply chain compromise
All this new-fangled stuff about transformers, LLMs, harnesses, agents, swarms of agents — and the infrastructure you need to have in place to actually build applications — is just the latest in a long line of developments.
As Sabretooth observed in the delightful cartoon animated series X-Men ‘97: “Key to surviving, keep up with technology.”
Telecommunications, clouds, cybersecurity, digital twins, identity and access management (IAM), cryptography, quantum cryptography … yada yada yada … AI.
What we’re getting at is that even if we don’t use a particular technology (we don’t claim to be practitioners), we can still competently write about it — to a level of depth that would probably surprise you. Plus, there’s an added bonus that we’re also able to relate these technologies to each other for a ‘rising tide’ effect.