Articles

posts

The answer isn’t complicated. We just need to peek under the hood of how LLMs actually work.

Seeing some panic in SEO/SEM circles over Google's Mode announcement. So, what now?

Generative models, like large language models, generate human-like text by predicting words based on patterns in datasets but struggle with negations, such as "not" in "I do not like apples." Training data favor affirmative statements, leaving fewer negated examples, so models poorly grasp how "not" inverts meaning, as seen in studies from Quanta Magazine and MIT Press Direct.

Why Generative Models Struggle with Negation

The market is buzzing with a flood of GTM agents that automatically identify target accounts. New AI solutions promise you can simply ask for a list of lookalike companies. Some even predict the GTM engineer job will disappear soon. If you just want a list of 50 companies from the Fortune 5000 list, all of the above is probably true. No need to read further.

20-50 Companies Max: The Uncomfortable Truth About AI-Generated Target Account Lists

The similarity gap, defined as the maximum gap in cosine similarity between a webpage and its nearest similar page, offers a lens to evaluate a website's uniqueness in the digital landscape. Cosine similarity measures how closely two pages align based on their content, typically represented as embeddings. A small similarity gap means a page closely resembles others, while a large gap signals distinctiveness. In our new AI-driven online world, standing out hinges on widening this gap, yet many websites inadvertently shrink it through common practices.

Understanding The Similarity Gap

Nuclear technology was once celebrated as the future of cheap, clean energy, “too cheap to meter.” Yet, it was pushed forward without enough safeguards. Toxic waste, for which we still have not found a permanent solution, and accidents like Chernobyl shattered public confidence. Even lower CO2 emissions have not been enough to restore its reputation.

From Nuclear Promises to AI Fallout: Are We Repeating the Same Mistakes?

The tech world is buzzing about AI coding assistants saving hours of developer time, but let's pause and ask the uncomfortable question: Are we just shifting the complexity from writing code to making our clients become unwitting debuggers of mysterious AI-generated solutions?

February 17, 2025

The Hidden Cost of AI-Generated Code: A Silicon Valley Reality Check

The evolution of B2B sales has taken another dramatic turn. We've moved from feature selling of the 1980s, through solution selling of the 1990s, and past the challenger/insight selling of the 2010s. Today, we're witnessing the rise of innovation-driven sales – and it's not just another methodology shift, it's a survival imperative.

February 12, 2025

Innovation Selling: The Last Frontier AI Can't Conquer in B2B Sales

You just upgraded your OpenAI account for the month to give Operator a shot and tasked it with cleaning up your CRM. Step one, an easy task to warm up: deduplicate and check if existing customers are already in the system. Surprisingly, you still find a few duplicates here and there, and some clients are still missing from the system.

February 12, 2025

Operators Are Standing By… But They Can’t Count!

The allure of instant creativity and wealth fueled by AI is undeniable, but are the current models sustainable? This post dives into the hidden costs of large generative AI models, from soaring hosting prices to immense energy consumption. It advocates for a more sensible approach, championing smaller, purpose-built AI tools that offer better performance and cost-effectiveness.

January 14, 2025

The Hype Cycle: Rideshare Dreams and AI Realities

We believe that sales and marketing should be about market penetration, not volume. By using precision company data and purpose-built LLMs, we’re enabling businesses to identify their TAM as a complete list of all Target Accounts.

January 13, 2025

Still going on fishing expeditions? Try TAM as an Account List instead.