Quartz 5

Future of Work

Properties1
tagsstrategy, automation, productivity

6 min read

Concept: Future of Work

The impact of artificial intelligence and automation on the global workforce, jobs, and the nature of human labor.

Key Principles

  • Strict Execution vs. Judgment: Tasks that require strict execution without judgment are easily automated, whereas roles involving strategic judgment, review, and direction are highly resilient.
  • Tiny Team Leverage: A three-person team equipped with advanced AI tools can now match the output volume of a 50-person traditional agency. While the output is not identical, it is close enough that clients question traditional agency overhead, shifting the competitive landscape toward speed, aggressive pricing, and portfolio quality (see tiny-team-leverage).
  • The Role of “Model Managers”: As AI models become more capable, a new class of human roles emerges—model managers. These individuals are responsible for the “care and feeding” of models: directing, aiming, feeding, judging, and reviewing their outputs.
  • Whole Job Delegation: Modern models like fable-5 represent a shift from short-lived prompt sessions to “whole job” execution (such as full consulting engagements). This requires humans to shift their focus from writing micro-prompts to defining high-level objectives, designing harnesses, and auditing business systems for high-complexity problems.
  • Human-in-the-Loop Gating: On sensitive, high-stakes domains (like health insurance appeals or tax filing), human judgment remains mandatory. Agents should be architected with strict gates—they prepare comprehensive “case files” with citations, but are restricted from final signing, submission, or payments. This shifts the focus of automation from button-clicking to cognitive preparation (see high-trust-agentic-work).
  • Redesigning the Building vs. Bolting on Motors: Merely running existing task lists through cheaper models is akin to “bolting an electric motor onto a steam-era factory central driveshaft” with minimal productivity gains. Real transformation requires redesigning team layouts, workflows, and review cycles around the speed of AI. (e.g., Stripe’s 50-million-line single-day code migration, which succeeded because they spent years building rigorous task coverage, automated verification, and review frameworks).
  • Manufacturing Imagination: Organizations cannot centralize AI strategy in a single hired visionary with no operational context. They must build a culture of distributed technical-imagination by putting domain experts who hold system context in touch with capable models and giving them the permission and resources to make bets.
  • The Society Transition Phase: The AI industry is moving beyond the initial capability innovation wave and entering a “society transition phase.” In this stage, the focal point shifts from raw performance metrics to the “return on AI in society,” centering on how intelligence is successfully integrated into existing regulatory, political, and corporate structures.
  • Enterprise Sticky Harnesses: Instead of selling basic API access, organizations must focus on building enterprise distribution systems. Deploying forward-deployed engineers and specialized integration tools (such as Anthropic’s “Claude Tag”) builds deep, custom-tailored harnesses around enterprise workflows, locking in recurring value as companies adapt and reinvent their organizational designs.
  • The Skill Evolution (Prompting -> Delegation -> Maintenance): The core human skill set has shifted from micro-prompting (articulating questions, 2023) to whole-job delegation (designing harnesses and hand-offs, 2025), and finally to persistent agent-maintenance (taking ownership, auditing running workflows, and decommissioning unowned systems, 2026).
  • The Shift to Agentic Org Design & Routing: Rather than single-model solutions, the future of work centers on orchestrating swarms under corporate-style hierarchies. Humans design agentic-org-design structures and routing rules, allowing highly capable and expensive models to manage while cheap models execute, achieving a 10x savings multiplier.
  • Verification-First Operations: Operational reliability shifts from trusting model outputs to designing continuous agent-verification-loops. Workforces manage by establishing strict standards (“Constitutions”) and letting checker agents run automated feedback loops, with humans acting as final strategic arbitrators.
  • Vague-to-Spec Translation: As code writing and pull request reviews are commoditized and automated by AI agents, the organizational center of gravity shifts to individuals who can translate vague business needs into precise, actionable specifications. Success is determined by human precision to direct machines and judgment to evaluate whether customer problems are solved.
  • The Coordination Trap: In leaner, AI-driven organizations, coordination roles are the first casualty. Professionals must migrate toward direct value creation (revenue-generating, customer-facing product, or data-driven business direction) to avoid this trap.
  • Adopting an Engineering Mindset: Knowledge work must adopt an engineering-mindset, characterized by precision, testability, falsifiability, and deep tool mastery to know when AI is wrong.
  • Metered Thinking & Budgeting Thought: Thinking is now metered and priced per token. The new managerial instinct is deciding which task in a week is worth purchasing thought for.
  • Velocity Safety: Cautious delay with AI is a high-risk strategy. Much like riding a bicycle, going slow makes balancing extremely difficult and increases the risk of falling behind. True stability and safety are achieved by leaning in, accelerating, and moving quickly to build experiential steadiness (see velocity-safety).
  • The AI-Native Operating System: High-velocity organizations are shifting from traditional roadmaps and meeting-heavy structures to an ai-native-operating-system. This framework focuses on moving repeatable human interactions toward code, using documentation-as-code as the primary active interface for AI agents, and requiring product managers and designers to work directly in the terminal and SDKs.
  • The ATS Arms Race: The traditional job application infrastructure has broken down due to an automated arms race between candidates using AI resume-gaming tools and employers deploying Applicant Tracking Systems (ATS). This dynamic has reduced cold application success rates to 0.4%, forcing a shift toward high-trust, alternative routing channels (see ats-arms-race).
  • Human Attention as the Non-Inflationary Resource: Applying Baumol’s Cost Effect and Jevons Paradox, as automated AI content generation becomes dirt-cheap, human attention and authentic human connection (“the string quartet”) become the ultimate scarce, non-inflationary assets. Platforms and creators must guard against “Claude-fishing” and low-effort slop to protect high-trust public discourse (see ai-slop).
  • Commoditization of Execution & Value Migration: As AI tools drop the cost of execution toward zero, outputs across feeds and markets homogenize (“samey results”). Because value does not disappear when execution gets cheap, it migrates to strategy, taste, spec precision, edge variance, and human judgment (see commoditization-of-execution).
  • Root-Cause Support Automation: Customer support in 2026 shifts from answering tickets faster at the end of the pipeline to automating non-linear context gathering across tools, maintaining human-approval gates for access and money, and eliminating underlying friction upstream (see root-cause-support-automation).
  • Overcoming AI Rollout Resistance: Successful corporate AI adoption requires explicit executive commitments to protect existing headcount and expand company horizons (as championed by jensen-huang). Rollouts must be scoped to specific bottom-line drivers with outcome metrics rather than token activity caps, transforming engineers into system designers and eval authors (see ai-rollout-resistance).
  • Dark Factory Software Engineering & Infinite Engineering Capacity: Autonomous software implementation pipelines do not replace visionary product thinkers, systems thinkers, and leaders who thrive under ambiguity. Instead, dark factories amplify them, turning a single product thinker into an operator with unlimited engineering capacity while shifting the core question from “can we build it?” to “should we build it?” (see dark-factory-software-engineering and these-are-the-people-ai-cant-replace).

References

Graph View

Backlinks