There's an uncomfortable truth in most organizations: institutional knowledge is power. The people who've been around longest know where the bodies are buried — which decisions were made and why, which approaches were tried and failed, which stakeholders care about what. New hires spend months building this context. Cross-functional partners never fully get it.
AI tools were supposed to democratize knowledge work. Instead, they've amplified the gap. Experienced team members know how to prompt effectively because they know the context. They can give their AI tool the right background. New hires can't, because they don't have the background to give.
The Onboarding Problem
We watched a new engineer join a team using SLEDS. On day one, they connected Claude to the team's sled. Immediately, they had access to every architectural decision, every design discussion, every open question — not as a static wiki, but as living context their AI tools could draw on.
Their first PR included a comment: "Based on the auth migration discussion in thread #47, I used the new Clerk flow instead of the legacy JWT approach." They'd been on the team for two days. Without SLEDS, discovering that decision would have required either reading months of Slack messages or asking someone who happened to remember.
Knowledge Shouldn't Be a Moat
When everyone's AI tools share the same context, the playing field levels. A junior developer's Claude session knows as much about the project as a senior architect's. A designer joining a cross-functional review has the same background as the engineers who built the feature.
This isn't about replacing expertise — architectural judgment and design intuition still matter. It's about separating expertise from context. You should get ahead because you think better, not because you happen to know things others don't have access to.
Making Knowledge Accessible
SLEDS threads aren't documentation in the traditional sense. They're not carefully curated wiki pages that go stale the moment they're published. They're living observations from real conversations, updated every time someone on the team discusses the topic. The context is always current because it's captured as a byproduct of actual work.
This is what shared memory enables: a team where context is infrastructure, not advantage. Where new hires contribute meaningfully in days, not months. Where cross-functional collaboration works because everyone's AI tools already understand the project.