Last quarter, we spent two months observing how teams use AI tools in practice. Not in demos or marketing videos — in actual daily work. The findings surprised us.
The Individual vs. Team Gap
Individual AI usage is high and getting higher. Most knowledge workers now use at least one AI tool daily. Many use two or three. But team-level AI usage — where the AI benefits compound across people — is still rare. Teams use AI tools individually, in parallel, without any shared context between them.
The high-performing teams we studied had figured out a workaround. They shared context manually: pasting conversation excerpts into Slack, maintaining shared prompt libraries, or having one person serve as the "AI coordinator" who fed context to everyone else's tools. It worked, but it was fragile and labor-intensive.
The Pattern
The common thread among effective AI teams was simple: they had found some way to make context shareable. Whether it was a shared Notion doc that everyone pasted into their AI tools, a Slack channel dedicated to "AI insights," or a team member who manually transferred context between conversations, the mechanism didn't matter. What mattered was that context moved between people and tools.
Teams without this pattern described a consistent frustration: "I had a great conversation with Claude about this, but I can't get that insight to my teammate without explaining the whole thing again." The AI tool was helpful in the moment but created knowledge silos.
What Good Looks Like
The best teams we observed treated AI conversations as team artifacts, not personal scratchpads. When someone had a productive session exploring a technical approach, the insights were captured and shared — not as a summary in Slack, but as context that other team members' AI tools could access.
This changes the dynamic of collaboration. Instead of five people having five separate conversations about the same topic, you get five people building on each other's AI-assisted thinking. Each conversation is informed by the prior ones. Ideas compound instead of being rediscovered.
Building for This
SLEDS was born from watching these patterns. The teams that figured out context sharing were 3-4x more productive with AI tools than teams that didn't. But the manual workarounds don't scale. You need infrastructure that captures and shares context automatically, across any tool, for any team member.
That's the product thesis: the value of AI tools increases non-linearly when context is shared. One person using AI is useful. A team sharing AI context is transformative.