Automation Analysis

AI Workforce Assessment

A rigorous analysis of 10 job functions, their automation potential, and what still requires human judgment.

10
Job Functions
68%
Avg. Automation
25+
Integrations
7
Workflows
SCROLL

Each discipline is a specialist

10 AI agents, each modeled after L5 Principal-level expertise. They share memory, communicate through an E8-encoded event bus, and reach consensus through Byzantine protocols.

๐Ÿ“‹
TPM
Program Manager
Planning Coordination

Orchestrates technical initiatives, manages dependencies, and forecasts risks. The strategic planner who sees the whole board.

Critical Path Monte Carlo Risk Forecast Status Reports Escalation
Automation Level 75%
๐ŸŽฏ
PM
Product
Planning Strategy

Product strategy, roadmaps, and prioritization. Uses RICE+EFE hybrid scoring to balance impact, confidence, and free energy.

Roadmaps RICE+EFE PRD Generation OKR Tracking Competitive Analysis
Automation Level 70%
๐Ÿ‘ฅ
EM
Manager
People Planning

Team health, capacity planning, and performance tracking. Predicts burnout while always requiring human judgment for people decisions.

Team Health Capacity 1:1 Prep Burnout Prediction Allocation
Automation Level 60%
Ceiling: ~75%
โš’๏ธ
ENG
Engineer
Technical Data

Code quality, architecture decisions, and tech debt. Reviews PRs with expert patterns and performs deep security analysis.

PR Review ADRs Security Scan Tech Debt Refactoring
Automation Level 80%
๐Ÿ”„
DE
Data
Technical Infrastructure

Data pipelines, quality monitoring, and schema management. Builds infrastructure for data flow with CDC, Great Expectations, and cost optimization.

Pipelines Data Quality Schema Registry CDC Cost Analysis
Automation Level 75%
๐Ÿ“Š
DS
Science
Research Analysis

ML experiments, causal inference, and fairness auditing. Uses DID, RDD, IV, and PSM for rigorous causal effect estimation.

Experiments Causal Inference Fairness Audit HPO Drift Detection
Automation Level 70%
๐Ÿ”
UXR
Research
Research Insights

User research, interview synthesis, and persona generation. Extracts themes from qualitative data at scale โ€” human facilitation required.

Interview Synthesis Personas Usability Tests Surveys Insights
Automation Level 55%
๐Ÿ“ฃ
PMM
Marketing
Marketing Strategy

Positioning, competitive intelligence, and launch planning. Generates battlecards with positioning jujutsu and real-time competitive alerts.

Positioning Battlecards Launch Plans Competitive Intel Messaging Tests
Automation Level 65%
๐ŸŽจ
Design
Designer
Creative Technical

UI/UX design, accessibility auditing, and design systems. Uses iterative design โ†’ critique โ†’ refine loops with multi-persona feedback.

Component Gen Accessibility Motion Specs Design System AI Critique
Automation Level 70%
๐Ÿ’Ž
QA
Quality
Quality Automation

Testing, chaos engineering, and safety verification. Has veto power on any action that violates safety constraints. The guardian of quality.

Chaos Engineering Test Generation Safety Verification Load Testing Mutation Testing
Automation Level 85%
Peak: Technical automation

Real workflows

Production integration workflows connecting Linear, GitHub, Notion, and Cursor. Each workflow shows how data flows between tools with automated triggers.

Linear โ†’ GitHub Sync Core Pipeline
๐Ÿ“‹ Linear Issue
โ†’
๐ŸŒฟ Create Branch
โ†’
๐Ÿ”€ Pull Request
โ†’
โœ… CI/CD
โ†’
๐Ÿ Close Issue

Linear cycles sync to GitHub milestones. PRs auto-update Linear issues. Merge closes linked issues.

GitHub โ†’ Linear Triggers Automated
๐Ÿ”€ PR Opened
โ†’
๐Ÿ“‹ Linear Issue
โ†’
โŒ CI Failed
โ†’
๐ŸŽซ Create Ticket
โ†’
๐Ÿ’ฌ Slack Alert

PR events trigger Linear issue creation. CI failures auto-create tickets with Slack notifications.

CI Failure โ†’ Cursor Fix AI-Assisted
โŒ CI Failure
โ†’
๐Ÿ“ GitHub Issue
โ†’
๐Ÿค– Cursor Agent
โ†’
๐Ÿ”€ Fix PR
โ†’
โœ… Auto-Merge

CI failures create GitHub issues with @cursor-bot. Cursor Background Agent fixes and auto-merges.

Research โ†’ Notion KB Knowledge
๐Ÿ” UXR Research
โ†’
๐Ÿ“š Notion DB
โ†’
๐ŸŽฏ PM Roadmap
โ†’
๐Ÿ“„ PRD Page
โ†’
๐Ÿ“‹ Linear Epic

Research findings store to Notion. PRDs link to Linear epics. Decision logs persist ADRs.

PR Review Pipeline Code Quality
๐Ÿ”€ PR Created
โ†’
๐Ÿค– CodeAnt AI
โ†’
โš’๏ธ ENG Review
โ†’
โœ… Approve
โ†’
๐Ÿ“‹ ADR to Notion

PRs get AI review + security scan (Snyk). Architecture decisions log to Notion as ADRs.

Sprint Sync TPM
๐Ÿ”„ Linear Cycle
โ†’
๐ŸŽฏ GitHub Milestone
โ†’
๐Ÿ“Š Notion Report
โ†’
๐Ÿ“ฃ Slack Summary

Cycles sync to milestones. Velocity tracked. Sprint reports generate to Notion + Slack.

25+ integrations

Each discipline connects to specialized external tools through type-safe contracts. From code analysis to competitive intelligence, the ecosystem extends capabilities far beyond the base model.

โš’๏ธ Code Quality
๐Ÿค– CodeAnt AI
๐Ÿ“Š SonarQube
๐Ÿ”’ Snyk
๐Ÿ›ก๏ธ Endor Labs
Used by: ENG
๐Ÿ’Ž QA Tools
๐Ÿค– Harness AI Test
๐Ÿ“ testRigor
๐ŸŒช๏ธ Harness Chaos
โšก k6 / Gatling
Used by: QA
๐ŸŽจ Design Tools
โ™ฟ Stark AI
๐ŸŽจ Figma AI
๐ŸŽจ Contrast Checker
Used by: Design
๐Ÿ”„ Data Tools
โš™๏ธ Dagster
๐Ÿ“š DataHub
๐Ÿ“ˆ Weights & Biases
๐Ÿงช Ax (Meta)
Used by: DE, DS
๐Ÿ“‹ Project Tools
๐ŸŽฏ nPlan
๐Ÿ”— Conductor
๐Ÿ—บ๏ธ Altirya
Used by: TPM, PM
๐Ÿ‘ฅ People Tools
๐Ÿ“ 15Five
๐Ÿ“Š Lattice
๐Ÿ’š Culture Amp
Used by: EM
๐Ÿ” Research Tools
๐ŸŽ™๏ธ Outset AI
๐Ÿ•Š๏ธ Dovetail
๐Ÿงช Optimal Workshop
Used by: UXR
๐Ÿ“ฃ Competitive Intel
๐Ÿ–๏ธ Crayon AI
๐ŸŒ Similarweb
โญ G2 Crowd
Used by: PMM

Automation levels

Current automation levels by discipline. Each percentage represents tasks that AI can perform todayโ€”the remaining work requires human judgment, creativity, or ethical oversight.

85%
QA
Test Gen, Chaos, Safety
Fully automated: Load tests, chaos experiments, coverage analysis, mutation testing
Human required: Test strategy, final release approval
80%
ENG
PR Review, Security, Analysis
Fully automated: PR review, security scanning, tech debt identification
Human required: Architecture decisions, production approvals
75%
TPM
Critical Path, Monte Carlo
Fully automated: Dependency graphs, simulations, status reports
Human required: Final escalations, stakeholder communication
75%
DE
Pipelines, Quality, Schemas
Fully automated: DAG generation, data quality checks, catalog mgmt
Human required: Platform architecture, data governance
70%
PM
Prioritization, PRDs, OKRs
Fully automated: RICE+EFE scoring, OKR tracking, epic decomposition
Human required: Strategic vision, customer discovery
70%
DS
HPO, Drift, Fairness
Fully automated: HPO, drift detection, feature management
Human required: Research design, business interpretation
70%
Design
Components, A11y, Systems
Fully automated: Design system enforcement, accessibility audit
Human required: Initial vision, design strategy
65%
PMM
Battlecards, Intel, Content
Fully automated: Content generation, competitive monitoring
Human required: Brand strategy, market intuition
60%
EM
Team Health, Capacity, 1:1s
Fully automated: Metrics aggregation, capacity calculation
Human required: ALL people decisions (hiring, promotions, coaching)
Ceiling: ~75% โ€” people decisions cannot be automated
55%
UXR
Synthesis, Personas, Surveys
Fully automated: Transcription, insight synthesis, survey design
Human required: Research planning, interview facilitation, rapport

Workforce Contraction

Based on current automation adoption rates, tech industry headcount will contract significantly by 2030. This visualization models three scenarios using S-curve adoption and verified 2025 data.

Base Headcount (2025)
5.74M
Jobs at Risk by 2030
600K
Reduction Rate
-10.5%
Wage Impact
$84B

Headcount by Discipline: 2025 vs 2030

ENG
3.57M (-13%)
PM
312K (-14%)
QA
211K (-23%)
Design
192K (-18%)
DE
151K (-18%)
EM
156K (-11%)
DS
130K (-16%)
TPM
95K (-17%)
PMM
74K (-16%)
UXR
36K (-14%)
2025 Baseline 2030 Projected
Data: CompTIA, BLS, LinkedIn Workforce Reports | Model: S-curve adoption
โš ๏ธ

Highest Risk: QA

40% current automation, 85% peak potential. AI test generation, load testing, and coverage analysis are production-ready today.

๐Ÿ›ก๏ธ

Most Protected: EM & UXR

People decisions and human research have the lowest ceilings. Hiring, promotions, and building rapport remain irreducibly human.

๐Ÿ“Š

The 30% Reality

Microsoft: AI generates ~30% of code. Copilot: 15M users, 51% faster coding. This is not speculationโ€”it's already here.

Full Model โ€” 7 Sheets

Task-level analysis, economic impact, YoY curves, and scenario comparison.

Assessment Framework

Each job function was analyzed for automation potential using a rigorous 3-tier classification system. Percentages are derived from implementation analysis, not speculation.

โœ“

Fully Automated

Tasks that can be performed entirely by AI without human intervention. Includes algorithmic processes, data aggregation, and pattern-based analysis.

  • RICE scoring calculations
  • PR review with CodeAnt AI
  • Data pipeline generation
  • Accessibility auditing
โ—

Partially Automated

Tasks where AI generates outputs but requires human review, validation, or approval before action. Includes LLM-generated content and recommendations.

  • PRD generation (needs requirements)
  • Architecture recommendations
  • Competitive analysis
  • Burnout prediction alerts
โœ—

Human Required

Tasks that fundamentally require human judgment, creativity, ethics, or relationship-building. These represent hard automation ceilings.

  • Strategic vision setting
  • All people decisions (hiring, promotions)
  • Customer interviews & rapport
  • Final deployment approvals
68%
Average Automation
85%
Highest (QA)
55%
Lowest (UXR)
75%
EM Ceiling