docs: add Neo data handling page - #20762
Open
workprentice[bot] wants to merge 2 commits into
Open
Conversation
Add content/docs/ai/neo/data-handling/_index.md, consolidating what's published about the models behind Pulumi Neo and how Neo handles the data it touches: which model family and provider power Neo, how access is scoped by the invoking identity's RBAC permissions, how prompts and task history are handled through Amazon Bedrock, and how secret values are redacted. Cross-links to the permissions model, security, and privacy pages, and links from the Neo landing page. Deliberately does not answer whether customer data is used for model training, retention duration, or model-change notification policy -- none of those are published anywhere today. See the PR description.
github-actions
Bot
added
review:triaging
Contributor
Pre-merge Review — Last updated 2026-08-07T16:30:32ZTip Summary: This PR adds a new docs page, Review confidence:
Investigation log
🔍 Verification trail52 claims extracted · 42 verified · 3 unverifiable · 0 contradicted
🚨 Outstanding in this PRNo outstanding findings.
|
github-actions
Bot
added
review:outstanding-issues
Collaborator
|
Your site preview for commit 2b83ede is ready! 🎉 http://www-testing-pulumi-docs-origin-pr-20762-2b83edef.s3-website.us-west-2.amazonaws.com Changed pages: |
Per pre-merge review: 'Amazon Bedrock does not retain prompts or completions' was an unqualified claim about Bedrock as a platform, but AWS's retention behavior is configurable and the guarantee that matters here is how Pulumi has configured its own Bedrock usage, not a universal property of Bedrock. Rephrase to make that scoping explicit.
Contributor
Author
|
@claude #update-review Fixed the outstanding finding: reworded the Bedrock retention sentence to scope it to Pulumi's own Bedrock configuration rather than asserting it as a universal property of Bedrock, using your suggested wording (2b83ede). On the low-confidence items: the runtime-session-teardown and context-rebuild sentences are drawn directly from the permissions page's existing 'Handling of secret values' bullet (already published, not new claims I'm introducing), so I've left them as-is. The model-update-cadence sentence is accurate first-party product description without a dedicated published citation, similar to other first-party statements elsewhere on this page and its siblings. |
github-actions
Bot
added
review:stale
Contributor
|
🤖 Review auto-refreshed — the latest push only touched lines with outstanding findings. |
github-actions
Bot
added
review:no-blockers
Contributor
|
Tagging in @kramhuber for a review before I merge. I'm sure you understand, @workprentice. |
Contributor
Author
|
This has been clean since the round-2 pre-merge review resolved the Bedrock-retention overclaim (no blockers, CI green, mergeable). Nudging for a maintainer merge -- nothing further pending on my end. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Adds a new page,
content/docs/ai/neo/data-handling/, that consolidates what's already published about the models behind Pulumi Neo and how Neo handles the data it touches: which model family and provider power it, how access is scoped by the invoking identity's RBAC permissions, how prompts and task history are handled through Amazon Bedrock, and how secret values are redacted. It cross-links to the permissions model,/security/, and/privacy/, and it's linked from two spots on the Neo landing page so it's discoverable without already knowing the URL.Why (fixes #20391)
A field person filed #20391 after being unable to answer a prospective customer's two questions from public docs: which model(s) power Neo (and whether that changes), and whether customer data is used to train or fine-tune models. The first question already has a published answer, just not gathered anywhere a reader would look for it — it's a sentence in the Neo landing page intro and a buried clause inside a "Handling of secret values" bullet on the permissions page. This PR surfaces that existing material as its own page instead of publishing anything new.
The reporter's original link (
/docs/pulumi-cloud/neo/) 301-redirects correctly to/docs/ai/neo/, so the "trouble finding the information" here is a discoverability gap, not a broken link.What this PR deliberately does not answer
Three things a trust-conscious reader would reasonably ask are not published anywhere today, and I'm not going to invent answers to close that gap:
The new page names these explicitly as open questions and points readers to their Pulumi account team rather than guessing. This needs product/legal sign-off before any of the three gets a firm published answer. If there's an existing internal answer to any of them, replacing that bullet with real copy is a trivial follow-up edit to this same page — happy to take a pass at that once the answer is confirmed.
Collision note
PR #20733 (Azure DevOps VCS integration docs refresh) is concurrently touching
content/docs/ai/neo/permissions/_index.md,code-reviews/_index.md, andpull-requests/_index.md. This PR does not edit any of those files — it only adds the newdata-handling/_index.mdand links to the permissions page rather than duplicating its content, and edits the Neo landing page_index.md, which #20733 does not touch.Verification
make lint— 0 errors (1849 files).valeon the changed/new files — 0 errors (2 pre-existing-stylewrite-goodwarnings, consistent with warnings already present elsewhere in this content).make build(yarn install,theme/stencil/infrastructureinstalls,make sync-icons,hugo --minify -e production) — 4,970 pages, zero template errors.aisidebar nav directly after "Permissions model" and before "Neo in the CLI" (menu weight 13), and that its internal links resolve.🧠 This PR was created by workprentice.