· The Skills Atlas
What dies.
Skills half life, 2000 to 2026.
A 6× decline in 26 years. The 2026 number is the one that breaks annual planning cycles. Quarterly is the new floor.
Median skills half life
Skills decay by category.
Where each capability falls on a five band scale. Green compounds. Red dies. The canonical view a Chief of Work (CoW), the executive who designs the work system across humans, agents, and robots, should review every quarter.
Compound or depreciate. There is no middle.
Train the left column. Stop overinvesting in the right. The CHRO's L&D budget should follow the green list, not the org chart.
Invest more every quarter.
- Judgment under ambiguityDecisions where stakes cannot be specified in a prompt.↑ career
- Taste and curationChoosing well from infinite AI options is the new bottleneck.↑ career
- Trust and persuasionBuyers, regulators, employees, partners. Stays human.↑ career
- OrchestrationDesigning systems of humans, agents, robots.↑ 10+ yr
- Learning velocityCompound is in employees who learn quickly.↑ 8+ yr
- Network capitalWho you know now eclipses what you know.↑ career
Stop training as a primary skill.
- Coding syntaxVibe coding compresses this to natural language.↓ 6 mo
- Tool memorizationEvery tool ships with a copilot. Adapters render this obsolete.↓ 10 mo
- Manual data entryAgents do this at $0.08 per task with higher accuracy.↓ 8 mo
- Routine analysisFirst draft analytical work is a prompt away from done.↓ 12 mo
- Standard documentationTemplates plus context yield production grade output.↓ 9 mo
- Procedural complianceRule based workflows are the first agent target.↓ 14 mo
For every skill, exactly one move.
No skill survives without an explicit decision. Pick one of four moves per capability. Redecide every quarter.
Compound
Invest aggressively. Judgement, taste, trust, orchestration. The list of skills that pay for themselves over a career.
Maintain
Domain expertise that ages slowly. Hold the line. Refresh annually. Do not overfund what is not decaying fast.
Compress
Skills the model has eaten. Move volume to agents under shadow oversight. Redirect human time to compound work.
Retire
Stop training new entrants. Convert remaining specialists into orchestrators. Close the discipline as a primary career path.
What the skills evidence points to.
Three takeaways recur across the Moonshots episodes that touch on skills, AI, and the workforce. Each is a reading of the evidence, not a prescription.
From annual planning to a continuous reading of skills.
Skills half life has collapsed
The episodes describe a fall in median skills half life from roughly 15 years around 2000 to about 2.5 years by 2026, a 6× decline. The headline figures are framed as estimates rather than precise measurements.
Compounding and depreciating skills are diverging
The same sources distinguish skills that compound over a career, such as judgement, taste, trust, and orchestration, from skills that depreciate fast as tools and agents absorb them, such as coding syntax and routine analysis.
Review cadence is shifting to continuous
A recurring forecast is that a half life measured in months sits awkwardly with annual review cycles, pushing skills planning toward shorter, more continuous reassessment. This is presented as a projection, not an observed outcome.
Four metrics. One quarterly pack.
The board readable view a Chief of Work delivers every 90 days. These four numbers, tracked together, replace traditional engagement scores in the workforce conversation.
Time to skill
Median time from identifying a capability gap to having it operationally covered. Down from 9 months in 2022.
Cost per unit work
Annual decline in blended cost of one operational output unit. The CFO readable metric.
Productivity per FTE
Output per fully loaded human equivalent. The unit shifts from person to person plus agent fleet.
Atlas coverage
Share of high cost role families with a deliberate compound / maintain / compress / retire designation.
drawn from the evidence.
Decay rates differ sharply by skill.
The heat map suggests a single training budget spread evenly across functions sits awkwardly with capabilities that decay at very different rates.
Compounding skills are the durable ones.
Judgement, orchestration, taste, and trust recur in the sources as skills that hold or grow in value while tool and syntax skills depreciate.
Cadence is the open question.
If half life is measured in months, the sources project that annual review cycles lag the underlying change. The right cadence is framed as an open question, not a settled answer.
One of five briefings distilled from Moonshots.
Drawn from over 250 episodes of Moonshots with Peter Diamandis and the Moonshot Mates. Figures are the sources' own estimates, and forecasts are noted as forecasts.