Skip to content
Agentic Levels

Everything starts here.

GuestLocal progress only
PreferencesSign in
01Start with one taskBest first move for beginners.02Check your LevelMeasure where you are.03Score an AI resultFind the habit to practice first.04Return to Your WorkScores, links, and checkpoints.
Start here

Begin

HomeThe main entry point.New to AIStart with one useful task.
Know where you are

Measure

Check your LevelUse this after you have tried AI.Fluency ScoreScore an AI result you can review.
Build the habit

Learn

LevelsLessonsTracks
Find the reference

Library

PromptsReferenceResourcesCompare Tools
Turn it into work

Apply

Your Next MoveChoose what AI should change next.Tool SetupGet the tools ready.
Come back later

Return

Your WorkScores, links, and checkpoints.My PathContinue from your level.Updates
Site

Site

PricingAboutFAQ & FeedbackPreferences

© 2026 Fuentes Studio

Privacy·Terms
yourCouncil
Ready to help
✦

What do you want to understand?

Ask anything about what you're learning.

Tracks›Chat With Your Knowledge
L3Lesson 1Free

Define the source boundary

What the system can and cannot know

After this, you'll be able to name the source set a knowledge assistant can use, what is excluded, and what answer types require human review.

Before you start

Complete Tool Crossover first if you are choosing between Projects, files, connectors, or a custom knowledge tool.

The idea

A knowledge workflow starts by naming what counts as source truth.

A user question reaches toward every possible company document with no source boundary.
A user question reaches toward every possible company document with no source boundary.
IncludedExcludedUnknownReview
MeaningValid sourceDo not useNot coveredHuman owner
ExampleHandbookDraft PDFMissing policyHR approval
RiskToo narrowStale answerGuessingSlow response

No source, no confident answer.

Boundary rule: the assistant can answer only from sources you load, connect, or cite. Anything outside that set is an unknown, not a guess.

The source boundary should include included folders, excluded folders, stale documents, private material, and answers that need human review.

Worked example: a team asks for policy answers. The source boundary includes the current handbook and HR updates, excludes old draft PDFs, and sends legal questions to HR before sharing.

Define the source boundary before asking questions.

Source boundaries prevent confident wrong answers. A knowledge assistant can sound like it knows the whole company when it only saw five files. The boundary makes that visible. It tells the tool what can be answered, what cannot be answered, and which answer types need a human owner.

Write the boundary in reader terms. "Use the current handbook and June HR update" is better than "use HR docs." "Do not answer compensation exceptions" is better than "be careful." The boundary should make refusal acceptable when the source set does not support an answer.

Try it (10 min)

Watch out for

  • Assuming the tool knows every company document.
  • Including stale drafts.
  • Letting private notes enter a team source set.
  • Skipping human review categories.

Paste this into Claude

Define the source boundary.

Knowledge task: [what people will ask]
Included sources: [folders, docs, files, links]
Excluded sources: [old, private, draft, unknown]
Human review needed for: [legal, HR, finance, client, other]

Return:
1. Included source set.
2. Exclusions.
3. Unknowns.
4. Human review rules.

Created by potrace 1.16, written by Peter Selinger 2001-2019 What a good response looks like

Included: current handbook, June HR update, approved FAQ. Excluded: old onboarding drafts and private manager notes. Unknowns: contractor policy after June. Review: HR approves legal or compensation answers.

Created by potrace 1.16, written by Peter Selinger 2001-2019 What good looks like

  • The boundary names included sources
  • It names exclusions
  • It keeps unknowns visible
  • It names human review rules

When this breaks

  • Breaks when excluded sources are silently treated as valid.
  • Breaks when answer scope is wider than the loaded source set.

AI can help with this

Use your knowledge assistant to help you you can name the source set a knowledge assistant can use, what is excluded, and what answer types require human review. Start with the exercise prompt and your real input. Ask for one draft, then check it against this proof: The boundary names included sources. Accept only the version you can verify yourself.

The source set separates into included, excluded, unknown, and human-review lanes.

Created by potrace 1.16, written by Peter Selinger 2001-2019 You can now

✓

You can name sources

  • ✓You can exclude stale material
  • ✓You can preserve unknowns
  • ✓You can require review

Key takeaways

Knowledge answers are only as trustworthy as the source boundary behind them.

  1. 1Included sources must be named.
  2. 2Excluded sources matter.
  3. 3Unknowns are allowed.
  4. 4Human review keeps risk bounded.

Created by potrace 1.16, written by Peter Selinger 2001-2019 Go deeper

  • Tool source boundary
  • ChatGPT Projects

Was this helpful?

Up nextPrepare the source packet→

Related lessons

Choose the connection typeBuild your first ProjectWhat Connectors are, and why they change everything
← Back to Chat With Your Knowledge