Build AI Harnesses to Make Every Employee a Power User
Top companies treat AI as growth tech, not efficiency tool, by creating institutional systems like Ramp's Glass that auto-configure workspaces, share 350+ skills via marketplace, and provide persistent memory—raising productivity floor for all from day one.
Leaders Capture 75% of AI Value Through Growth and Reinvention
PwC study shows 20% of companies claim 75% of AI economic gains by using AI 2-3x more for identifying growth opportunities and 2.6x more for reinventing business models, rather than mere productivity. McKinsey's analysis of 20 AI leaders across industries confirms: AI drives 20% EBITDA uplift, breaks even in 1-2 years, and generates $3 incremental EBITDA per $1 invested—focusing on economic leverage points like supply chain in automotive, not efficiency alone. Leaders build enduring capabilities around tech, upskill senior business leaders (not just IT), keep 70%+ AI talent in-house, treat platforms as strategic assets fed by ongoing data enrichment, and master agentic engineering: ingesting unstructured data, adding agents with automated guardrails, and codifying playbooks from experiments. Speed defines advantage as skills half-life shortens.
Institutional AI Solves Coordination and Scaling Challenges
George Sulka's a16z essay argues individual AI boosts personal productivity 10x, but no company is 10x more valuable without institutional AI—distinct processes aligning individual efforts. Key pillars: coordination prevents chaos from uncoordinated AI outputs (e.g., varying prompts creating misaligned work); signal extraction amid content explosion; professional objectivity over individual alignment. Institutional AI scales revenue, not just time saved, by directing agents/humans via clear OKRs, roles. Without this, AI amplifies quirks like cloned super-employees rowing oppositely.
Ramp's Glass Blueprint: Auto-Enable Power Users Organization-Wide
Ramp built Glass after only 9% daily AI use due to painful setups; now every employee gets a pre-configured workspace via SSO integrating tools like Ramp CLI, Salesforce, Gong, Slack, Notion. Core principles: (1) Preserve full upside—hide complexity (multi-window workflows, automations, persistent memory) without dumbing down, as AI tutors help users tackle hard problems. (2) Propagate breakthroughs—one team's skill becomes team's baseline via Dojo marketplace (350+ reusable agent skills, e.g., Zendesk workflow pulling tickets/account health for resolutions). (3) Product as enablement—Sensei AI recommends top 5 role-relevant skills from usage history. Features: Day-one memory synthesis from connections/projects (daily pipeline updates via Slack/calendar); scheduled cron automations posting to Slack even offline. Build in-house for moat (internal productivity edge), speed (same-day fixes via Slack triage), and product insights (translates to customer finance tools). Result: New hires start productive; users learn by doing as features implicitly teach (skills show great outputs, memory proves context value). This harness engineering at org scale compounds advantages competitors outsourcing can't match.