Raise the Floor vs Raise the Ceiling
Summary
AI adoption in organizations operates on two distinct axes: raising the floor (making everyone more capable) and raising the ceiling (pushing the frontier of what’s possible for power users). These require different tools, different change management strategies, and different success metrics — but both are necessary for meaningful organizational transformation.
Raising the floor means bringing baseline AI capability to everyone in the org — including people who would never write a prompt from scratch. Tools like Google Gems, Custom GPTs, and Claude Projects are ideal here: packaged, shareable, low-friction runtimes that encode expert judgment into reusable AI tools anyone can use. The change management motion is distribution and adoption — get pre-built tools into people’s hands, lower the barrier to first use, measure breadth of adoption.
Raising the ceiling means enabling power users and early adopters to push into genuinely new workflows — agentic methods, multi-tool orchestration, AI-native development patterns. This is where Claude Code, Cursor, MCP integrations, and custom agent architectures live. The change management motion is enablement and experimentation — create space for exploration, share learnings back, measure depth of capability.
The risk is over-indexing on one axis. Floor-only strategies create a veneer of adoption without transformative impact. Ceiling-only strategies create pockets of excellence that don’t spread. Effective AI change management sequences and balances both.
How to Apply
When to use: When designing an AI adoption strategy for a team or organization. When evaluating whether current efforts are balanced. When deciding which AI tools and initiatives to invest in.
The diagnostic:
- What are we doing to raise the floor? (Gems, packaged GPTs, shared templates, no-code AI tools)
- What are we doing to raise the ceiling? (Agentic tooling, custom workflows, power-user enablement)
- Are these connected? (Do ceiling insights get packaged into floor tools? Do floor users have a path to ceiling capabilities?)
Sequencing considerations:
- Floor-first is often the right starting move — it builds organizational comfort with AI and creates demand for more
- Ceiling work generates the insights and patterns that become the next generation of floor tools
- The two should feed each other: power users discover patterns → patterns get packaged into runtimes → runtimes raise the floor → new baseline creates demand for the next ceiling push
Tool mapping:
- Floor tools: Google Gems, Custom GPTs, Claude Projects, Zapier AI workflows, pre-built templates
- Ceiling tools: Claude Code, Cursor, Codex, MCP integrations, custom agent architectures, agentic workflows
- Bridge tools: Shared prompt libraries, skill files, documented workflows that translate ceiling discoveries into floor-accessible patterns
Origin
Context: Observation from thinking about AI adoption patterns in non-AI-native organizations — the tension between broad shallow adoption and deep narrow adoption. How you use it: As a lens for evaluating whether an AI adoption strategy is balanced and self-reinforcing, and for classifying AI tools by which axis they primarily serve. Why it works: It names the two distinct motions that “AI adoption” actually requires, prevents conflating them, and creates a framework for sequencing investments.
If you later find external sources that discuss this idea, add a Sources section alongside this Origin section and update origin to “both”.
Related
- Scale Manager Expertise With AI — Raising the floor: packaging expert judgment into AI tools others can use
- Reverse Engineer Judgment Into AI — The mechanism for converting ceiling expertise into floor tools
- Tool Identity as Adoption Gate — Floor tools fail when identity/positioning filters out intended users
- Adjacent Disciplines — Adoption Dynamics — The floor/ceiling dynamic plays out differently across collaborator disciplines