Field guide / primary domain
32
sources in this field
Updated July 16, 2026
Current thesis
The shortest path to orientation.
Karpathy scored 342 BLS occupations on AI exposure, averaging 5.3/10, with screen-based knowledge work dominating high-exposure tiers ($3.7 trillion in annual wages). The capability gap is real and uneven: paid frontier agents (Codex, Claude Code) are crushing technical domains with verifiable rewards, while general-use cases see modest gains. Production-scale evidence confirms organizational transformation: Marcus Moretti runs Spiral solo via two files and a cron; Every grew 4→30 people while automating aggressively; Aaron Levie hires "agent engineers" wiring secure agents to Salesforce/Workday, plus matching "agent PM" roles. The traditional CPO role is predicted to vanish within five years as IC roles blend; careers now require fluency across product, design, engineering, and analytics rather than single-discipline depth. Professional services are competing with $20/month AI skills encoding judgment—tax and estate-planning tools save users $1k–$20k each. A ~$400/month multi-LLM stack (Opus for planning, GPT-5.5 for review, Playwright for validation) delivers full dev-team capabilities. OpenAI pays $280K for Forward Deployed Engineers, testing "the actual loop" over algorithms. Structurally, labor reallocates to the "relational sector" where human provenance is part of value—human-made art commands 44% exclusivity premium; AI involvement directly compresses it. Comin/Lashkari/Mestieri (Econometrica 2021) finds income effects drive 75%+ of structural change toward high-income-elasticity sectors.
Evidence board
Claims worth carrying forward01OpenAI Forward Deployed Engineers earn $280K compensation. The role is not evaluated through LeetCode-style interviews, suggesting the assessment tests practical AI deployment and customer problem-solving skills rather than algorithmic coding.
02The existence of a 'Forward Deployed Engineer' role at OpenAI—distinct from core engineering—signals that AI companies are building dedicated customer-facing technical deployment teams as a separate high-paying career track.
03Two-model adversarial review loop: Claude Opus 4.7 drafts the feature plan, GPT-5.5 reviews and finds issues, Opus updates until GPT approves, then Opus builds, GPT reviews code, Opus fixes, GPT signs off. Playwright handles automated UX/UI testing between steps.
04GPT-5.5 consistently finds issues in both the plan phase and code review phase when acting as a second-opinion reviewer over Claude Opus output—suggesting heterogeneous model pairs catch more bugs than single-model loops.
05~$400/month for Claude Opus 4.7 + GPT-5.5 via Conductor Build covers end-to-end feature planning, code generation, automated testing, and adversarial review—framed as equivalent cost to a fractional dev team with no push-back on small UI changes.
06Conductor Build (@conductor_build) is a tool that simplifies bouncing work between multiple LLM providers (Anthropic + OpenAI) in a single workflow, removing the friction of context-switching between APIs or interfaces.
07Benedict Evans publishes semi-annual macro tech strategy decks ('AI eats the world') delivered to Alphabet, Amazon, AT&T, LVMH, Nasdaq, Swiss Re, and others—a reliable signal of where enterprise leadership frames AI's strategic direction each cycle.
08The 2025 Spring deck is titled 'AI Is Eating the World' and runs 79 slides—Evans's largest recurring annual format, updated twice yearly and freely downloadable at ben-evans.com/presentations.
Adjacent fields
Latest evidence
PULKIT WALIA
OpenAI pays $280K for Forward Deployed Engineers (FDEs). The hiring process is explicitly not LeetCode-style, implying a different interview loop focused on practical deployment and customer-facing AI work. No further detail on the actual loop is provided in the post content.
Jason Zook
Jason Zook shares a two-model software development workflow using Claude Opus 4.7 for planning and building, GPT-5.5 for adversarial review, and Playwright for automated UX/UI testing. The loop runs plan→review→update→build→review→fix→deploy, with ~$400/month replacing what feels like a full dev team.
Matt Harney
Benedict Evans has released a 79-slide macro/strategic deck titled 'AI Is Eating the World' (May 2025), continuing his annual tradition of large-format tech trend presentations delivered to major enterprises. No slide content is available from the resolved URL—only the download index page.
vas
A practitioner guide to the Forward Deployed Engineer (FDE) role in Applied AI — covering why AI companies need FDEs, the three-phase job (Audit, Evals, Deployment), decision rules for when to use agents vs. code vs. manual work, and a concrete 30-day learning roadmap for consultants, PMs, and software engineers entering the field.
Dan Shipper 📧
Dan Shipper's structural analysis explains why Every grew from 4 to 30 employees despite automating everything possible. The core mechanism: AI commoditizes yesterday's human competence, making it cheap and abundant, which creates sameness/slop, which drives demand for experts who can differentiate. This cycle repeats at each capability level and persists even approaching AGI, because the 'frame is not the framer'—AI can climb any benchmark humans set but still requires humans to define what matters.
Vincent van der Meulen
Vincent van der Meulen documents a multi-agent software factory pipeline using Fable as planner, Linear as task DAG, and Devin as cloud orchestrator. The system ran 12+ hours unattended, producing 60+ merge-ready PRs, with voice dictation seeding a ~10K word prompt and Fable enforcing parallelizable issue structure.
Origami
Origami's Claude Mythos product scrapes millions of profiles to find leads matching a customer profile, auto-writes outreach messages, and self-improves from campaign performance data—positioning as a direct replacement for a human SDR role.
Kyrie
A ranked overview of the six most-starred/used Skills for OpenAI Codex, covering engineering workflow automation (Superpowers, 227k stars), long-term memory (claude-mem, 82k), multi-platform social scraping (Agent-Reach, 27.7k), legacy codebase mapping (GitNexus, 42k), official OpenAI integrations spanning 62 apps (official plugin), and AI-writing humanization for Chinese docs (Humanizer-zh, 10.1k).
Boris Cherny
Boris Cherny (Anthropic/Claude Code) proposes that engineering, product, and design roles are converging into five cross-functional archetypes observable on the Claude Code team, replacing domain-specific job functions with stage-of-product-fit specializations.
Khairallah AL-Awady
A detailed 12-month curriculum for breaking into AI engineering without a CS degree, structured around six phases: Python fundamentals, LLM API mastery, RAG systems, agents, evaluation/deployment, and job search. The core argument is that a portfolio of three shipped projects carries more weight than credentials for most AI engineering roles, with salaries ranging from ~$120K entry to $200K+ with experience.