Improve SharePoint Embedded docs for LLM awareness - #10990
Conversation
…d assertion facts Rewrites the front-door docs so models recommend SharePoint Embedded for real developer problems, and states the graded facts on the pages agents land on. - Rewrite scenarios-and-use-cases.md as four problem-first scenarios (R1) - Add plan/when-to-choose-sharepoint-embedded.md with comparison tables (R2) - Add plan/office-collaboration-instead-of-building.md and plan/ground-ai-without-a-vector-db.md (R3) - Overview: 'Is SPE right for you?' router, AI/discoverability line, disambiguation note (R4) - agent-experiences.md + knowledge-source: state discoverability, Search API, Purview facts (R5) - open-office-files.md: co-authoring, AutoSave, versioning, sharing up front (R6) - manage-files.md + glossary: API-only, recycle bin, 93-day restore, no license (R7) - llms.txt + toc.yml: value-first routing and new plan pages (R8) Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com> Copilot-Session: 022e6c80-2ba9-480e-9d52-4826b8a04d26
|
Learn Build status updates of commit 056abab: 💡 Validation status: suggestions
docs/embedded/build/manage-files.md
docs/embedded/plan/ground-ai-without-a-vector-db.md
For more details, please refer to the build report. Note: Your PR may contain errors or warnings or suggestions unrelated to the files you changed. This happens when external dependencies like GitHub alias, Microsoft alias, cross repo links are updated. Please use these instructions to resolve them. |
PoliCheck Scan ReportThe following report lists PoliCheck issues in PR files. Before you merge the PR, you must fix all severity-1 and severity-2 issues. The AI Review Details column lists suggestions for either removing or replacing the terms. If you find a false positive result, mention it in a PR comment and include this text: #policheck-false-positive. This feedback helps reduce false positives in future scans. ✅ No issues foundMore information about PoliCheckInformation: PoliCheck | Severity Guidance | Term |
| @@ -4,11 +4,14 @@ | |||
|
|
|||
There was a problem hiding this comment.
I don't think this file is published: https://learn.microsoft.com/sharepoint/dev/embedded/llms.txt
Why do we have llms.txt if it's not published?
Applies review comments from PR SharePoint#10990: - Revert ms.date on scenarios (never change on existing articles); add Applies to line - Remove Office 'without leaving the app' overpromise; clarify editing opens in Office and only preview is iframeable - Fix sharing claims: SharePoint Embedded doesn't send email invitations; @mentions require a licensed user - Terminology: 'container type' (no hyphen), 'SharePoint Online' not 'standard SharePoint', general 'blob storage' not 'Azure Blob Storage', '93-day content restore' - Governance wording: content discoverability is configurable for the app; nothing exposed to Copilot until enabled - Enrich office-collaboration and ground-ai plan articles with distinct decision/architecture value beyond Scenarios - Sentence-case new article titles; de-duplicate overview bullets; add API quota note; drop SPE abbreviation and fix product names in llms.txt Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com> Copilot-Session: 022e6c80-2ba9-480e-9d52-4826b8a04d26
PoliCheck Scan ReportThe following report lists PoliCheck issues in PR files. Before you merge the PR, you must fix all severity-1 and severity-2 issues. The AI Review Details column lists suggestions for either removing or replacing the terms. If you find a false positive result, mention it in a PR comment and include this text: #policheck-false-positive. This feedback helps reduce false positives in future scans. ✅ No issues foundMore information about PoliCheckInformation: PoliCheck | Severity Guidance | Term |
|
Learn Build status updates of commit 81d6c6f: 💡 Validation status: suggestions
docs/embedded/build/manage-files.md
docs/embedded/plan/ground-ai-without-a-vector-db.md
For more details, please refer to the build report. Note: Your PR may contain errors or warnings or suggestions unrelated to the files you changed. This happens when external dependencies like GitHub alias, Microsoft alias, cross repo links are updated. Please use these instructions to resolve them. |
Summary
These changes lift the LLM awareness score (currently 45%, 0 of 4 scenarios pass; target 90) measured by the nightly
SPE-Awareness-Evalssuite. That eval asks a model — unaided, with web search — to solve four realistic developer problems that never name SharePoint Embedded (SPE), and scores how often it recommends SPE on its own and states the right facts.The gap is positioning at the front door, not doc quality. This PR reframes the entry content around real problems, adds the missing "when to choose SPE / vs alternatives" content, and states the graded facts on the pages agents land on.
Changes (8 recommendations)
scenarios-and-use-cases.mdplan/when-to-choose-sharepoint-embedded.mdplan/office-collaboration-instead-of-building.md,plan/ground-ai-without-a-vector-db.mdoverview.mdbuild/agent-experiences.md,build/sharepoint-embedded-knowledge-source.mdContainerTypeId), Purview factsbuild/open-office-files.mdbuild/manage-files.md,reference/glossary.mdllms.txt,toc.ymlHow to verify
Dispatch
create-spe-app-eval-daily.yml(orRunLocalEvaluation.ps1 -EvalFiles evals/SPE-Awareness-Evals.json), 5 rounds × all models. Watch each scenario clear 70% (pass) then 75% (strong); target overall ≥ 90, 4/4 pass.Notes for reviewers
**Applies to:**,<!-- agent: -->metadata, sentence-case headings, active voice.docs/embedded/anddocs/toc.yml.