# Use AI Clearly

> Task-first AI education and deterministic decision support.

## Canonical sections

### Learn

- [Context versus model knowledge](https://useaiclearly.com/learn/context-vs-model-knowledge/): Understand context versus model knowledge through practical distinctions, a worked example, and checks you can apply to a real task.
- [Cost per completed task](https://useaiclearly.com/learn/cost-per-completed-task/): Understand cost per completed task through practical distinctions, a worked example, and checks you can apply to a real task.
- [Current facts and source checking](https://useaiclearly.com/learn/current-facts-source-checking/): Understand current facts and source checking through practical distinctions, a worked example, and checks you can apply to a real task.
- [Data sensitivity and approved environments](https://useaiclearly.com/learn/data-sensitivity-approved-environments/): Understand data sensitivity and approved environments through practical distinctions, a worked example, and checks you can apply to a real task.
- [Grounding and retrieval](https://useaiclearly.com/learn/grounding-and-retrieval/): Understand grounding and retrieval through practical distinctions, a worked example, and checks you can apply to a real task.
- [Hallucinations and unsupported claims](https://useaiclearly.com/learn/hallucinations-unsupported-claims/): Understand hallucinations and unsupported claims through practical distinctions, a worked example, and checks you can apply to a real task.
- [Inherent task risk versus safeguards](https://useaiclearly.com/learn/inherent-risk-and-safeguards/): Understand inherent task risk versus safeguards through practical distinctions, a worked example, and checks you can apply to a real task.
- [What an AI model can and cannot know](https://useaiclearly.com/learn/model-knowledge-limits/): Understand what an AI model can and cannot know through practical distinctions, a worked example, and checks you can apply to a real task.

### Solution patterns

- [Bounded agent](https://useaiclearly.com/patterns/bounded-agent/): Learn when bounded agent is the simplest fit, where it fails, and which controls set its safe boundary.
- [Conventional software](https://useaiclearly.com/patterns/conventional-software/): Learn when conventional software is the simplest fit, where it fails, and which controls set its safe boundary.
- [Direct prompt](https://useaiclearly.com/patterns/direct-prompt/): Learn when direct prompt is the simplest fit, where it fails, and which controls set its safe boundary.
- [Human-only](https://useaiclearly.com/patterns/human-only/): Learn when human-only is the simplest fit, where it fails, and which controls set its safe boundary.
- [Prompt with context](https://useaiclearly.com/patterns/prompt-with-context/): Learn when prompt with context is the simplest fit, where it fails, and which controls set its safe boundary.
- [Retrieval-grounded workflow](https://useaiclearly.com/patterns/retrieval-grounded/): Learn when retrieval-grounded workflow is the simplest fit, where it fails, and which controls set its safe boundary.
- [Tool workflow](https://useaiclearly.com/patterns/tool-workflow/): Learn when tool workflow is the simplest fit, where it fails, and which controls set its safe boundary.

### Scenarios

- [Answering questions about current policy](https://useaiclearly.com/scenarios/policy-question-answering/): See how answering questions about current policy is assessed, which rules drive the result, and what the next safe step requires.
- [Answering questions over internal documents](https://useaiclearly.com/scenarios/internal-document-qa/): See how answering questions over internal documents is assessed, which rules drive the result, and what the next safe step requires.
- [Building a recurring report](https://useaiclearly.com/scenarios/recurring-report/): See how building a recurring report is assessed, which rules drive the result, and what the next safe step requires.
- [Calculating invoice totals](https://useaiclearly.com/scenarios/invoice-calculation/): See how calculating invoice totals is assessed, which rules drive the result, and what the next safe step requires.
- [Choosing between rules and an LLM](https://useaiclearly.com/scenarios/choose-rules-or-llm/): See how choosing between rules and an LLM is assessed, which rules drive the result, and what the next safe step requires.
- [Classifying support requests](https://useaiclearly.com/scenarios/classify-support-requests/): See how classifying support requests is assessed, which rules drive the result, and what the next safe step requires.
- [Creating medical or financial guidance](https://useaiclearly.com/scenarios/medical-financial-guidance/): See how creating medical or financial guidance is assessed, which rules drive the result, and what the next safe step requires.
- [Drafting an internal email](https://useaiclearly.com/scenarios/draft-internal-email/): See how drafting an internal email is assessed, which rules drive the result, and what the next safe step requires.
- [Extracting structured data from documents](https://useaiclearly.com/scenarios/extract-document-data/): See how extracting structured data from documents is assessed, which rules drive the result, and what the next safe step requires.
- [Generating marketing variations](https://useaiclearly.com/scenarios/marketing-variations/): See how generating marketing variations is assessed, which rules drive the result, and what the next safe step requires.
- [Producing customer-facing claims](https://useaiclearly.com/scenarios/produce-customer-claims/): See how producing customer-facing claims is assessed, which rules drive the result, and what the next safe step requires.
- [Publishing scheduled social posts](https://useaiclearly.com/scenarios/publish-social-posts/): See how publishing scheduled social posts is assessed, which rules drive the result, and what the next safe step requires.
- [Reconciling duplicate records](https://useaiclearly.com/scenarios/reconcile-duplicate-records/): See how reconciling duplicate records is assessed, which rules drive the result, and what the next safe step requires.
- [Researching current market information](https://useaiclearly.com/scenarios/research-current-market-information/): See how researching current market information is assessed, which rules drive the result, and what the next safe step requires.
- [Reviewing a contract](https://useaiclearly.com/scenarios/review-contract/): See how reviewing a contract is assessed, which rules drive the result, and what the next safe step requires.
- [Running a multi-step browser workflow](https://useaiclearly.com/scenarios/multi-step-browser-workflow/): See how running a multi-step browser workflow is assessed, which rules drive the result, and what the next safe step requires.
- [Screening job applicants](https://useaiclearly.com/scenarios/screening-job-applicants/): See how screening job applicants is assessed, which rules drive the result, and what the next safe step requires.
- [Summarizing sensitive support records](https://useaiclearly.com/scenarios/summarize-sensitive-records/): See how summarizing sensitive support records is assessed, which rules drive the result, and what the next safe step requires.
- [Summarizing supplied meeting notes](https://useaiclearly.com/scenarios/summarize-meeting-notes/): See how summarizing supplied meeting notes is assessed, which rules drive the result, and what the next safe step requires.
- [Updating a CRM](https://useaiclearly.com/scenarios/update-crm/): See how updating a CRM is assessed, which rules drive the result, and what the next safe step requires.

### Workflows

- [Bounded actions](https://useaiclearly.com/workflows/bounded-actions/): Apply bounded actions with explicit prerequisites, evidence, review, and stop conditions.
- [Document extraction with validation](https://useaiclearly.com/workflows/document-extraction/): Apply document extraction with validation with explicit prerequisites, evidence, review, and stop conditions.
- [Draft and review](https://useaiclearly.com/workflows/draft-and-review/): Apply draft and review with explicit prerequisites, evidence, review, and stop conditions.
- [Irreversible actions](https://useaiclearly.com/workflows/irreversible-actions/): Apply irreversible actions with explicit prerequisites, evidence, review, and stop conditions.
- [Research with sources](https://useaiclearly.com/workflows/research-with-sources/): Apply research with sources with explicit prerequisites, evidence, review, and stop conditions.

### Verify

- [Citation checking](https://useaiclearly.com/verify/citation-checking/): Turn citation checking into an operational review with concrete evidence, checks, and stop conditions.
- [Evaluation sets and acceptance criteria](https://useaiclearly.com/verify/evaluation-sets-acceptance-criteria/): Turn evaluation sets and acceptance criteria into an operational review with concrete evidence, checks, and stop conditions.
- [Output checks](https://useaiclearly.com/verify/output-checks/): Turn output checks into an operational review with concrete evidence, checks, and stop conditions.
- [Qualified workflow review and escalation](https://useaiclearly.com/verify/qualified-review/): Turn qualified workflow review and escalation into an operational review with concrete evidence, checks, and stop conditions.
- [Source traceability](https://useaiclearly.com/verify/source-traceability/): Turn source traceability into an operational review with concrete evidence, checks, and stop conditions.
- [Verifiability](https://useaiclearly.com/verify/verifiability/): Turn verifiability into an operational review with concrete evidence, checks, and stop conditions.

### Troubleshoot

- [Stale facts](https://useaiclearly.com/troubleshoot/stale-facts/): Diagnose stale facts, correct the workflow, and prove the original failure is controlled.
- [Unsafe actions](https://useaiclearly.com/troubleshoot/unsafe-actions/): Diagnose unsafe actions, correct the workflow, and prove the original failure is controlled.
- [Unsupported claims](https://useaiclearly.com/troubleshoot/unsupported-claims/): Diagnose unsupported claims, correct the workflow, and prove the original failure is controlled.

### Reference

- [Approved data environments](https://useaiclearly.com/reference/data-environments/): Use this plain-language reference for approved data environments, including its operational test and limits.
- [Autonomy levels](https://useaiclearly.com/reference/autonomy/): Use this plain-language reference for autonomy levels, including its operational test and limits.
- [Disallowed uses](https://useaiclearly.com/reference/disallowed-uses/): Use this plain-language reference for disallowed uses, including its operational test and limits.
- [High-impact decisions](https://useaiclearly.com/reference/high-impact-decisions/): Use this plain-language reference for high-impact decisions, including its operational test and limits.
- [Inherent risk bands](https://useaiclearly.com/reference/risk-band/): Use this plain-language reference for inherent risk bands, including its operational test and limits.

### Glossary

- [Acceptance criteria](https://useaiclearly.com/glossary/acceptance-criteria/): Acceptance criteria explained with a practical test, an example, and its effect on an AI workflow.
- [Agent](https://useaiclearly.com/glossary/agent/): Agent explained with a practical test, an example, and its effect on an AI workflow.
- [Approved environment](https://useaiclearly.com/glossary/approved-environment/): Approved environment explained with a practical test, an example, and its effect on an AI workflow.
- [Audit trail](https://useaiclearly.com/glossary/audit-trail/): Audit trail explained with a practical test, an example, and its effect on an AI workflow.
- [Autonomy cap](https://useaiclearly.com/glossary/autonomy-cap/): Autonomy cap explained with a practical test, an example, and its effect on an AI workflow.
- [Bounded action](https://useaiclearly.com/glossary/bounded-action/): Bounded action explained with a practical test, an example, and its effect on an AI workflow.
- [Citation checking](https://useaiclearly.com/glossary/citation-checking/): Citation checking explained with a practical test, an example, and its effect on an AI workflow.
- [Context](https://useaiclearly.com/glossary/context/): Context explained with a practical test, an example, and its effect on an AI workflow.
- [Control](https://useaiclearly.com/glossary/control/): Control explained with a practical test, an example, and its effect on an AI workflow.
- [Data minimization](https://useaiclearly.com/glossary/data-minimization/): Data minimization explained with a practical test, an example, and its effect on an AI workflow.
- [Disposition](https://useaiclearly.com/glossary/disposition/): Disposition explained with a practical test, an example, and its effect on an AI workflow.
- [Escalation trigger](https://useaiclearly.com/glossary/escalation-trigger/): Escalation trigger explained with a practical test, an example, and its effect on an AI workflow.
- [Evaluation set](https://useaiclearly.com/glossary/evaluation-set/): Evaluation set explained with a practical test, an example, and its effect on an AI workflow.
- [External action](https://useaiclearly.com/glossary/external-action/): External action explained with a practical test, an example, and its effect on an AI workflow.
- [Fit band](https://useaiclearly.com/glossary/fit-band/): Fit band explained with a practical test, an example, and its effect on an AI workflow.
- [Grounding](https://useaiclearly.com/glossary/grounding/): Grounding explained with a practical test, an example, and its effect on an AI workflow.
- [Hallucination](https://useaiclearly.com/glossary/hallucination/): Hallucination explained with a practical test, an example, and its effect on an AI workflow.
- [Human check](https://useaiclearly.com/glossary/human-check/): Human check explained with a practical test, an example, and its effect on an AI workflow.
- [Inherent risk](https://useaiclearly.com/glossary/inherent-risk/): Inherent risk explained with a practical test, an example, and its effect on an AI workflow.
- [Least privilege](https://useaiclearly.com/glossary/least-privilege/): Least privilege explained with a practical test, an example, and its effect on an AI workflow.
- [Model](https://useaiclearly.com/glossary/model/): Model explained with a practical test, an example, and its effect on an AI workflow.
- [Pilot](https://useaiclearly.com/glossary/pilot/): Pilot explained with a practical test, an example, and its effect on an AI workflow.
- [Prompt injection](https://useaiclearly.com/glossary/prompt-injection/): Prompt injection explained with a practical test, an example, and its effect on an AI workflow.
- [Qualified reviewer](https://useaiclearly.com/glossary/qualified-reviewer/): Qualified reviewer explained with a practical test, an example, and its effect on an AI workflow.
- [Residual risk](https://useaiclearly.com/glossary/residual-risk/): Residual risk explained with a practical test, an example, and its effect on an AI workflow.
- [Retrieval](https://useaiclearly.com/glossary/retrieval/): Retrieval explained with a practical test, an example, and its effect on an AI workflow.
- [Rollback](https://useaiclearly.com/glossary/rollback/): Rollback explained with a practical test, an example, and its effect on an AI workflow.
- [Source quality](https://useaiclearly.com/glossary/source-quality/): Source quality explained with a practical test, an example, and its effect on an AI workflow.
- [Source traceability](https://useaiclearly.com/glossary/source-traceability/): Source traceability explained with a practical test, an example, and its effect on an AI workflow.
- [Stop condition](https://useaiclearly.com/glossary/stop-condition/): Stop condition explained with a practical test, an example, and its effect on an AI workflow.
- [Verifiability](https://useaiclearly.com/glossary/verifiability/): Verifiability explained with a practical test, an example, and its effect on an AI workflow.

## Methodology and policies

- [Methodology](https://useaiclearly.com/methodology/)
- [Editorial ownership and review](https://useaiclearly.com/about/editorial-review/)
- [Editorial policy](https://useaiclearly.com/about/editorial-policy/)
- [Source policy](https://useaiclearly.com/about/source-policy/)
- [Corrections](https://useaiclearly.com/about/corrections/)
- [Commercial neutrality](https://useaiclearly.com/about/commercial-neutrality/)
- [Ownership and referrals](https://useaiclearly.com/about/ownership/)
- [Privacy](https://useaiclearly.com/privacy/)