"Where Do You Fall on the Eight Levels of AI Adoption?" — Mike Taylor

Why this is in the vault

A clean eight-rung maturity ladder for AI adoption (chatbot through manager-of-sub-agents) that gives RDCO a shared vocabulary to locate exactly where Ray-the-COO-agent sits and which rung the unhobbling thrust is climbing toward.

The core argument

Taylor's framing-correction up front: chasing the viral power-user ("12 Claude Code sessions in parallel") is the wrong target. The right level for any given task is set by how much you trust the AI to do the job unsupervised and how costly a mistake would be — a higher rung is not automatically better.

The ladder, each rung defined by who does the work and where the human sits:

1. Chatbot — you ask, it answers.
2. Copilot — AI works alongside you, inside your files.
3. Agent — executes a task step by step, checking in for approval.
4. Autopilot — runs on its own; you review the result after.
5. Workflows — you build a system that makes the output more reliable.
6. Assistant — works in the background without being prompted.
7. Multi-agent — you manage several long-running agents at once.
8. Orchestrator — a manager agent runs a team of sub-agents for you.

The guide pairs each rung with sample prompts, what becomes possible at that stage, and signals for when it is time to move up. The through-line: match the level to the trust-and-blast-radius of the work, and periodically audit whether you are leaving value on the table by staying too low.

Note on house self-promo: the post's footer is Every's standard product-and-consulting block — it plugs Spiral, Sparkle, Cora, Monologue, and Proof, plus Every's "AI training, adoption, and innovation" consulting service. The author, Mike Taylor, is Every's head of tech consulting, so the eight-levels framework also functions as a soft top-of-funnel for that consulting arm. The article itself is not a product pitch; the framework stands on its own.

Mapping against Ray Data Co

Strong, and unusually precise — this ladder is almost a self-assessment built for RDCO.

• Where Ray-the-COO already sits: rungs 6–8, not 4. The always-on Mac mini agent works in the background without being prompted (cron loops: morning-prep, open-threads-check, sync-contacts, curiosity), which is Taylor's Level 6 (Assistant). The sub-agent fan-out pattern (one subagent per newsletter article, per research question, per critic axis) plus the manager-runs-sub-agents shape is Level 8 (Orchestrator) by his definition. RDCO is not climbing toward multi-agent — it is already operating there on specific surfaces.
• The L4→L5 tension restated in his vocabulary. The founder places RDCO at L4 and is building toward L5 (see project memory + [[2026-05-06-dec-ai-not-replacing-curious-developers]]). Taylor's rungs reframe that gap: the unhobbling work is largely Level 5 (Workflows) labor — building systems that make agent output reliable (verification-as-independent-worker, fresh-eyes critic subagents, the 12-stage production workflow). The honest read is that RDCO has reached high rungs on capability (6–8) while the reliability scaffolding (rung 5) is the actual bottleneck. That inverts the naive "we need to get more autonomous" instinct — the next unlock is reliability, not more autonomy.
• The trust-and-blast-radius criterion is already encoded as hard gates. Taylor's "how big a deal if it messes up" is exactly why RDCO keeps deploy/production-writes and external email sends behind human gates while letting reversible work (Notion adds, vault notes, paper-trade drafts) run autonomously. The ladder gives a clean external articulation of the IC-mode vs production-mode split.
• phData agent-activation tie-in. The day-job (phData) is selling organizational agent adoption; this ladder is a ready-made client-facing diagnostic — meet a team where they are on the eight rungs, then sequence them up. Useful framing artifact, not just personal reference.

Honest caveat: the ladder is a vocabulary, not a measurement. It flatters by giving RDCO high-rung labels; the real question (rung-5 reliability) is the unglamorous one the framework underweights.

Related

• [[2026-05-21-every-after-automation]] — Dan Shipper on what comes after the automation rungs; same Every house view on orchestration.
• [[2026-05-06-dec-ai-not-replacing-curious-developers]] — the L4→L5 unhobbling thesis this ladder reframes (developer-as-supervisor).
• [[2026-05-15-every-team-agents-vs-personal-pets]] — Every's own employees climbing these rungs in practice.
• [[2026-05-26-every-codex-for-knowledge-work-power-user-guide]] — adjacent Every guide on pushing knowledge work up the ladder.
• [[2026-01-20-every-ai-teaching-management]] — prior Mike Taylor piece; managing agents as a management skill.