skip to content

$ weekly notes on building & governing AI agents, in your inbox subscribe →

about experiments advisory speaking tools frameworks blog contact

$

→ about → experiments → advisory → speaking → tools → frameworks → blog → contact $ subscribe to newsletter

›

assessment

Is your team agent-ready?

Seven questions. Sixty seconds. A tier mapping for where you sit and which engagement fits best. No email required, no calendar dump. No flattering yourself.

Your answers stay in your browser. Nothing is sent anywhere.

$ question_1_of_7

How many agents (defined: persistent role + tools + audit trail) does your team operate in production today?

[A] Zero. We're in proof-of-concept land. [B] 1–2 pilots, one engineer can name them. [C] 3–10. We have a registry. [D] 10+. We have a team that owns them.

$ question_2_of_7

Who can name your eval set off the top of their head?

[A] We don't have one. [B] One person, in a spreadsheet. [C] It's checked into git, but rarely run. [D] It runs on every change. We track the trend.

$ question_3_of_7

If a model output hallucinates a customer attribute today, what happens?

[A] We'd never know. [B] The user might flag it. [C] We have alerting on a downstream signal. [D] We have pre-output validation + an audit log.

$ question_4_of_7

Who owns the responsible-AI policy at your org?

[A] Nobody, officially. [B] Legal, but it hasn't been updated this year. [C] A working group across legal + AI + product. [D] A named function with a budget and a roadmap.

$ question_5_of_7

How does a new MCP server / agent tool get approved for production?

[A] Anyone installs anything. [B] Slack-based vibe check. [C] Documented review, sometimes followed. [D] Mandatory review + scope audit + sandbox period.

$ question_6_of_7

How are agent skills versioned and shared across teams?

[A] They aren't. Copy-paste prompts. [B] Internal wiki. [C] A skills registry, used inconsistently. [D] A skill platform with semver and consumers.

$ question_7_of_7

If your CFO asked tomorrow what your AI agents have cost — and saved — what would you say?

[A] I'd ballpark and pray. [B] I'd pull invoices and guess outcomes. [C] We track cost. Outcomes are anecdotal. [D] Both, by team, by quarter, by use case.

0 / 7 answered

result

best fit:

$ explore engagement →

// work

/tools /skills /frameworks /advisory /experiments /education /speaking /portfolio /process /assessment /faq /contact

// meta

/about /now /uses /tags /elsewhere

// writing

/newsletter /guides /essays /downloads /constellation /lab

// elsewhere

github ↗ linkedin ↗ marketing + media alliance ↗ ai leadership think tank ↗ /rss

$ last_build: 2026-08-10 · © 2026 alec foster

github linkedin rss

> esc

↑↓ navigate ↵ select esc close

$ keyboard_shortcuts.md esc

// navigation

⌘K or ctrl+K — open command palette
? — toggle this overlay
esc — close any overlay
/ — jump to command palette input

// command palette

↑ ↓ — navigate results
↵ — select / navigate
type a tag like #agents — search by tag
type a service like workshop — find related pages

// 404 page

cd <path> — navigate
ls — list pages
help — show commands
clear — clear history

// agent swarm

click anywhere — spawn a task
drag a rectangle — dispatch a team
hover over a worker — see its model

based on conventions from cmd+k, vercel, linear, and most claude code workflows