Skip to main content

Raise the Floor

One person's breakthrough should become everyone's baseline. The organizations that figure this out will compound faster than anyone playing solo.

The Pattern​

Most organizations adopt AI the same way: individually. Someone on the team discovers a workflow. They get faster, smarter, more productive. But their discovery stays locked inside their setup. Everyone else is still figuring it out from scratch.

This is the solo player trap. Everyone is leveling up independently, grinding the same learning curve, making the same mistakes, and reaching the same breakthroughs that their colleague reached three weeks ago. The organization's AI capability is the sum of individual efforts, not the compound of shared knowledge.

Raising the floor is the alternative. It means building infrastructure where every individual discovery gets captured, packaged, and distributed so that the minimum capability of every person in the organization rises with each breakthrough.

The goal is not to lower the ceiling for power users. It is to raise the floor for everyone else.

Why This Is Different from Other Concepts​

This is not permissionless knowledge. Permissionless knowledge is about scaling one expert's insight to many people through courses, docs, and automated delivery. Raise the Floor is about the organizational flywheel where discoveries flow laterally, peer to peer, through shared infrastructure.

This is not self-improving enterprise. Self-improving enterprise is about the business itself evolving through feedback loops and automated improvements. Raise the Floor is more specific: it is about the mechanism by which human capability compounds across a group.

This is not compounding docs. Compounding docs is about documentation getting more useful as it grows. Raise the Floor is about people getting more capable because someone else's work is now their starting point.

The distinction matters because the failure mode is specific: most organizations have smart people doing smart things in isolation. The fix is not more training. It is shared infrastructure.

What Raising the Floor Looks Like​

Shared Skills​

The clearest implementation is a shared skill library. When someone figures out the best way to accomplish a task with AI, they encode it as a reusable instruction file (a markdown skill file) and publish it where others can install it.

Ramp built this pattern at scale with their internal tool Glass. They created a marketplace called the Dojo where employees publish skills that any colleague can install. Over 350 skills have been shared company-wide. A sales rep packages a Gong analysis workflow. A CX engineer packages a Zendesk investigation process. Through the Dojo, entire teams level up overnight. (See the full Ramp case study.)

The Applied AI Society does the same thing at the community level. Every concept, playbook, and framework on this site is a shared skill in a different format. When a practitioner discovers a better way to scope a client engagement, it becomes a playbook. When a chapter leader figures out how to run a great event, it becomes a guide. The community's floor rises with each contribution.

Propagated Best Practices​

Beyond explicit skills, raising the floor means propagating the implicit knowledge that makes people effective. What does "good" look like? What questions should you ask before starting a project? What mistakes do people commonly make?

This is why Virgil Abloh's "Free Game" was revolutionary. He was not just sharing techniques. He was sharing the operating system behind his creative process: how he sourced materials, how he thought about design, how he built brands. The details that people normally gatekeep because they feel like competitive advantage. Virgil understood that in a world where knowledge flows freely, the advantage goes to the community that shares best, not the individual who hoards most.

The same logic applies to AI adoption. The companies where one person's configuration discovery becomes a company-wide default will outpace companies where everyone reinvents the wheel.

Product as Enablement​

Ramp's key insight was that the product itself teaches faster than any training session. When someone installs a skill and immediately gets a useful result, they learn what good AI output looks like without attending a workshop. When memory auto-populates from their connected tools, they learn that context is the difference between a generic answer and a useful one.

This is the encounter at the organizational level. You do not convince people that AI is valuable through presentations. You hand them a tool that already works because someone else's breakthrough is baked into the default experience.

The Flywheel​

Raise the Floor compounds in a specific way:

  1. Someone discovers a workflow. A sales rep finds a way to use AI to prep for calls in half the time.
  2. They package it as a skill. The workflow becomes a reusable instruction file.
  3. The organization distributes it. Every sales rep can now install the skill.
  4. The floor rises. The minimum capability of the sales team just increased.
  5. New discoveries happen from the higher floor. Because everyone is starting from a better baseline, the next breakthroughs are more sophisticated.
  6. Repeat.

Each cycle raises the floor higher. The organization's capability compounds not just from individual effort but from the network effect of shared knowledge. This is what it looks like when an organization becomes a self-improving enterprise at the human level, not just the systems level.

How to Start​

If you are a team lead or manager:

  1. Create a shared folder (or repo) for skill files. Even a Google Drive with markdown files works to start.
  2. When someone on your team discovers a useful AI workflow, ask them to write it down as a skill file. Ten minutes of documentation saves dozens of hours across the team.
  3. Make skill sharing part of team rituals. A five-minute "skill of the week" in your standup goes further than a quarterly AI training session.

If you are a practitioner working with clients:

  1. Every engagement should produce at least one reusable skill file. The client's specific workflow encoded as an instruction file.
  2. With permission, generalize successful patterns into skills that other clients can benefit from. Your library grows with every engagement.

If you are building a community:

  1. Open-source your best knowledge. The Applied AI Society docs exist because the highest-leverage thing we can do is raise the floor for everyone, not sell access to a ceiling.
  2. Create structures for members to contribute. The floor only rises when discoveries flow in both directions.

The ceiling takes care of itself. Power users will always push the boundaries. The strategic question is: how fast is your floor rising?

Further Reading​

  • Ramp: Glass: The corporate case study. 700 employees, 350 shared skills, and the Dojo marketplace that made it compound.
  • Instruction Files: The unit of shareable knowledge. Skill files are how breakthroughs get packaged.
  • Permissionless Knowledge: The related pattern for scaling one expert's knowledge to many. Raise the Floor is the organizational, peer-to-peer version.
  • The Self-Improving Enterprise: Where the flywheel leads at the systems level.
  • The Encounter: Why hands-on experience teaches faster than workshops. The product is the enablement.
  • Compounding Docs: Every shared skill makes every other skill more useful. The knowledge graph compounds.
  • Your Two Futures: The fork every person and organization faces. Raising the floor is how organizations choose Future A.