Everyone Feels Behind

Hi there,

There is one thing that is true across every team we talk to, regardless of industry, size, or how far along they are:

everyone feels behind.

The flood of announcements, new products, new features, and the general noise of the space is overwhelming even for those of us who do this for a living.

Not long ago, if your team was tinkering with Claude or ChatGPT, you were probably ahead of the curve. If you had built a few Skills and trained them to write in your voice, you were doing well. Based on what we are seeing, these behaviors are no longer enough. The bar is going up.

At Legible we talk to a lot of teams about where they are in their AI journey. A few patterns are emerging from the ones that are keeping up.

People management. The most advanced companies have done a leveling exercise on their team and have a plan to bring everyone up. They are identifying the builders inside their organizations and giving them more room: bigger token budgets, fewer constraints. The automatic backfill is over. When someone leaves the first question is whether AI can pick up the work. And they are being deliberate about the AI dividend. When your team saves five hours a week, where do those hours go? Does workload stay flat, go up, or go down? They are thinking hard about the difficult question of: what is in it for the employee?

Context management. The most advanced teams are not just enabling MCP to all of their systems and letting everyone scrape what they can find. They are being intentional about where their business context lives. Whether that is Google Drive, Notion, or GitHub, the goal is a shared source of truth that agents can actually work with cleanly. And they are keeping their tool footprint deliberately small. Every additional tool is an attack surface.

Token management. The best teams are building dashboards to understand where tokens are going and mapping consumption to teams and initiatives. The old BI dashboards reported what happened. AI dashboards need to tell you what to do about it.

Manifestos. Do not sleep on the manifesto. The best companies are treating their AI manifesto as a living document, not a one-time exercise. Expected behaviors, competencies, values: all of it needs to evolve as the tools do. It’s helpful to have a vision of where you are trying to get to.

Mental models. The best teams are moving beyond the ChatGPT or Claude license rollout. They are experimenting with agents that live in Slack, that text with them, that have a defined role and a job to do. Skydive, Grokbot, Instinct, Muse, Viktor all offer different ways to work with AI.

We are hosting an event in a couple of weeks on exactly this topic. If any of the above resonates, grab a spot.

I read every reply. Let me know what I missed.

Lawrence