EDITOR’S LIST / ORDERED BY TASK STAGE

Read less. Choose what changes the next step.

This editorial path selects 26 guides by task stage. The full catalog lists all 63 published guides.

Choose an entry point ↓Browse all guides →

AI VISTA / READING FILE

EDITORIAL ORDER01—04

01Make the job concrete4 GUIDES
02Make output reviewable6 GUIDES
03Connect evidence and freshness3 GUIDES
04Control automated consequences13 GUIDES

ENTER WHERE YOU AREAI VISTA

01ONE DECISION FIRST

CURATED BY EDITORSNO POPULARITY WEIGHTONE JOB PER SET

LEARNING ROUTE / STAGE 04 · READING LIST

This stage fills the current decision gap without letting reading replace practice.

The editorial order follows four work stages: framing, review, evidence, and operations. Find the stage you are in, then read one or two guides that change the next action.

BRING: One unresolved decision from the course
DO HERE: Choose one or two guides by work stage
LEAVE WITH: A corrected decision or practical tool

WHEN THIS STAGE IS DONE Record a practice note Name the conditions, failure, or question to test; the note stays in this browser.

BEFORE YOU READ

Write three sentences before opening a guide.

Without a decision, an artifact, and evidence, another saved article usually becomes another unfinished task. Name all three so reading becomes a testable revision.

STOP RULEStop once one guide lets you edit the artifact and schedule the next test. Return only when a new failure reveals a new decision gap.

01
DECISION TO CHANGE
What decision am I willing to revise after reading?
For example: whether support triage may write directly to the ticketing system.
02
ARTIFACT TO EDIT
Which piece of working material will this reading change?
For example: failure handling in the task brief or the approval level in a permission ladder.
03
EVIDENCE OF VALUE
What observable change would make the reading worthwhile?
For example: fewer misrouted cases in a ten-case pilot while risky cases still reach a person.
→
READY TO READ
All three sentences point to the same real task.
READY

A COMPLETE READING CHOICE

Not ‘this looks interesting,’ but ‘this can change the material on my desk.’

This example chooses one task-brief guide and stops as soon as it produces a testable revision. The list exists to shorten the distance between a blocker and the next practice—not to extend reading time.

Turn an AI idea into a testable task brief →

AI VISTA / READING TRACE01 GUIDE SELECTED

01
CURRENT PROBLEM
The team wants AI to handle the support inbox, but people disagree about which actions ‘handle’ includes.
↓
02
WHY THIS GUIDE
The gap is not another tool. User, input, acceptable output, boundary, and handoff are not in one working definition.
↓
03
REVISION AFTER READING
Replace ‘handle email’ with ‘suggest queue and urgency; send low-confidence and mixed-intent cases to the support lead,’ then complete one task brief.
↓
04
WHEN TO STOP
When two colleagues can use the brief to reach the same continue, revise, or stop judgment on ten cases, run the pilot instead of opening another guide.
■

EDITORIAL INCLUSION & EXCLUSIONOPEN CRITERIA

01
INCLUDE: it changes one concrete decision
A guide states the task moment in which it helps and which working artifact should change afterward.
02
INCLUDE: it has cases, an artifact, and a boundary
The method survives at least one uncomfortable case and provides a usable tool plus conditions where it does not fit.
03
EXCLUDE: concept-only, ranking, or generalized claims
Popularity, unsupported effect numbers, generic model rankings, and tool inventories do not enter the editorial order.

FRAME

Make the job concrete

For turning an AI idea into testable work and choosing who controls the next step.

4 GUIDES

01 Should this task use AI at all?Judge a task by variation, verifiability, stakes, data boundaries, and fallback before choosing a model or building a workflow. Task framing 02 Turn an AI idea into a testable task briefReplace a vague AI feature request with a one-page brief that names the user, input, acceptable output, evidence, boundaries, and fallback. Task framing 03 Workflow or AI agent? Choose the control loop before buildingDecide when a fixed AI workflow is enough and when an agent needs bounded control, tool feedback, stop rules, and a human handoff. Task framing 04 Run a ten-case AI pilot before you scaleBuild a deliberately small test set with ordinary, difficult, ambiguous, and unsafe cases so a promising demo becomes evidence. Evaluation

REVIEW

Make output reviewable

For variable output and unclear failure causes.

6 GUIDES

01 Map the context before rewriting the promptSeparate instructions, source material, examples, history, tools, and output rules to see what the model actually receives. Context and prompts 02 When should you compact a long AI conversation?Set a compaction checkpoint from measured context use, preserve decisions and unresolved actions, then test whether the next turn still follows the facts. Context and prompts 03 Write the acceptance contract before the promptDefine required fields, evidence, tone, uncertainty, and rejection conditions before optimizing prompt wording. Context and prompts 04 Choose JSON Schema, Markdown, or plain text for AI outputMatch an AI response format to its downstream consumer, then route refusals, truncation, parser errors, schema errors, and factual errors separately. Context and prompts 05 Test prompt injection in documents and tool outputsBuild a regression set for untrusted pages, files, retrieval snippets, and tool results; check intended behavior and permission gates before release. Context and prompts 06 Diagnose an AI answer before changing the modelTrace a bad answer to missing input, conflicting instructions, weak evidence, capability limits, or a broken handoff. Evaluation

GROUND

Connect evidence and freshness

For answers that require citations, access, and version control.

3 GUIDES

01 Decide whether RAG is worth buildingCompare knowledge volatility, source volume, citation needs, access rules, and maintenance cost before adding retrieval. Grounded answers 02 Evaluate RAG retrieval before judging the answerBuild a small retrieval trial that checks gold passages, no-answer questions, document versions, ranking, and access boundaries before generation. Grounded answers 03 Test citations for support and freshnessAudit whether each important claim is supported by the cited passage, uses the right version, and survives a freshness check. Grounded answers

OPERATE

Control automated consequences

For agents that use tools, keep long-term memory, or change external state.

13 GUIDES

01 Give an agent the smallest useful permissionPlace each action on a five-level permission ladder, from read-only suggestions to explicitly approved irreversible work. Agent safety 02 Design an AI agent tool contract before exposing the functionSpecify an agent tool's input schema, preconditions, authorization, side effects, idempotency, timeout results, and trace before making it callable. Agent architecture 03 When should an AI agent continue, retry, or stop?Use observable progress, separate turn and cost budgets, error classes, and duplicate-action checks to govern each step of an agent run. Agent architecture 04 Subagent or multi-agent? Choose the delegation topologyCompare one agent, a manager calling specialists, a conversation handoff, and parallel delegation by control ownership, shared state, cost, and failure recovery. Agent architecture 05 When should an AI agent save, correct, or delete long-term memory?Design a cross-session memory rule for what to save, how to handle a changed preference, and how to verify deletion across retrieval and retained copies. Agent architecture 06 Design recovery before the agent failsWrite the stop signal, saved state, owner, rollback action, and safe retry rule before automation reaches production. Agent safety 07 Choose a model with a job-specific scorecardWeight quality, latency, tool use, privacy, recovery, and operating constraints using cases from the workflow you will ship. Model operations 08 How should an AI model service fail over? Plan retries, queues, and human handoffRoute a failed model request by error type, user deadline, data boundary, and an evaluated backup model instead of treating every outage as a retry. Model operations 09 How much VRAM does a local LLM need? Budget weights and KV cacheEstimate raw model weights and KV cache separately, test context length and concurrency, then measure the missing peak costs on your serving stack. Local model deployment 10 How many concurrent requests can a local LLM serve?Find the highest tested arrival rate that meets first-token, output-speed, queue, error, and memory targets on your actual inference server. Local model deployment 11 Local LLM or cloud API? Choose by data, peak load, and operationsCompare owned hardware, a team-managed rented GPU, and a managed model API using data paths, measured peak capacity, operating ownership, and cost per accepted task. Local model deployment 12 Map the full cost of an AI workflowCount input, output, retries, tools, waiting, review, and failure recovery instead of comparing a single token price. Model operations 13 Why prompt caching misses: run a three-request testMeasure cache writes, reads, latency, and accepted-task cost with a stable prefix and a changed-prefix control, using provider-specific rules. Model operations

Courses and topics open directly from the main navigation.

Courses

Topics

Read less. Choose what changes the next step.

This stage fills the current decision gap without letting reading replace practice.

Write three sentences before opening a guide.

What decision am I willing to revise after reading?

Which piece of working material will this reading change?

What observable change would make the reading worthwhile?

All three sentences point to the same real task.

Not ‘this looks interesting,’ but ‘this can change the material on my desk.’

INCLUDE: it changes one concrete decision

INCLUDE: it has cases, an artifact, and a boundary

EXCLUDE: concept-only, ranking, or generalized claims

Make the job concrete

Make output reviewable

Connect evidence and freshness

Control automated consequences