The Track
A Section Blog

The hidden reasons you’re not getting AI ROI

AI is good enough, the humans need help
Newsflash if you’re waiting for AI to get better before you invest: It’s already pretty good, it’s the people using it that need to get a lot better – and fast. Greg is taking over this week’s newsletter to show you why.

When to use a specialized AI tool vs. an LLM
If you’re hitting the limits of your LLM because you need more data or niche training, you’re likely doing a lot of high-end knowledge work – and that’s where specialized AI tools come into play.

We tested two Deep Research tools. One was unusable.
Two LLMs have released Deep Research features without much of a splash. So if you’ve been wondering about them (and don’t want to pay the $200/month price tag), read on for our Chief of Staff’s take on ChatGPT vs. Gemini.

How an agency is preparing for AI clients
AI has put a big question mark over the future of agencies. So we talked to David Freas, Managing Partner at Supermoon, about how he’s transforming his brand agency with AI agent teammates – and for AI agent clients.

Can you build a team of just AI tools?
The same question has persisted all year: Can you reduce headcount by just using AI tools? Google’s Amit Rawal is trialing a largely AI-based team, and here are his lessons.

How to redesign roles with AI
Augmenting workflows is just the beginning of working with AI. To unlock huge time savings and strategy opportunities, eventually you need to rethink those processes altogether.

Squishy vs. hard ROI: Why leaders need both
One of the hardest parts of AI ROI is figuring out how to report qualitative wins to execs who want quantitative reports. At our AI:ROI Conference, Michael Domanic shared the framework for approaching these two disparate kinds of ROI.

AI’s progress may be holding your team back
On paper, the rapid release of new AI models and features looks like a win for knowledge workers. In reality, teams are drowning in AI overwhelm. Here’s how leaders can help them scale an ever-steeper learning curve.




