forgejo_admin/westside-ai-assistant
1
0
Fork 0
Code Issues 13 Pull requests Projects Releases Packages Wiki Activity Actions

Upgrade Nemo model from Sonnet 4 to Sonnet 4.6 #28

New issue
Open
opened 2026-03-31 01:08:06 +00:00 by forgejo_admin · 3 comments
forgejo_admin commented 2026-03-31 01:08:06 +00:00
Owner
Copy link

Type

Feature

Lineage

Standalone — improves SQL generation quality for the db_query tool (#27).

Repo

forgejo_admin/pal-e-deployments

User Story

As Marcus (admin)
I want Nemo to understand complex questions and generate accurate SQL
So that I get correct answers the first time

Context

Nemo currently runs claude-sonnet-4-20250514 (set in pal-e-deployments/overlays/westside-ai-assistant/prod/deployment-patch.yaml line 16). Upgrading to claude-sonnet-4-6-20250514 provides better SQL generation, schema reasoning, and natural language understanding. The k8s deployment env var ANTHROPIC_MODEL defaults to claude-sonnet-4-20250514 in the base but is overridden by the kustomize patch.

One-line change in the deployment patch YAML.

File Targets

Files to modify:

  • overlays/westside-ai-assistant/prod/deployment-patch.yaml line 16 — change claude-sonnet-4-20250514 to claude-sonnet-4-6-20250514

Files NOT to touch:

  • Application code — model is read from env var
  • Base deployment — override stays in the overlay

Acceptance Criteria

  • ANTHROPIC_MODEL env var set to claude-sonnet-4-6-20250514 in kustomize overlay
  • Nemo responds correctly after ArgoCD sync
  • Latency acceptable (under 5s for typical questions)

Test Expectations

  • Manual: ask Nemo a question, verify response quality
  • Manual: verify deployed env var matches

Constraints

  • Persist in kustomize overlay (not kubectl set env)
  • Monitor Anthropic API costs after upgrade

Checklist

  • PR opened
  • Tests pass
  • No unrelated changes

Related

  • project-westside-basketball
  • westside-ai-assistant #27 (db_query tool — benefits from better model)
### Type Feature ### Lineage Standalone — improves SQL generation quality for the db_query tool (#27). ### Repo `forgejo_admin/pal-e-deployments` ### User Story As Marcus (admin) I want Nemo to understand complex questions and generate accurate SQL So that I get correct answers the first time ### Context Nemo currently runs `claude-sonnet-4-20250514` (set in `pal-e-deployments/overlays/westside-ai-assistant/prod/deployment-patch.yaml` line 16). Upgrading to `claude-sonnet-4-6-20250514` provides better SQL generation, schema reasoning, and natural language understanding. The k8s deployment env var `ANTHROPIC_MODEL` defaults to `claude-sonnet-4-20250514` in the base but is overridden by the kustomize patch. One-line change in the deployment patch YAML. ### File Targets Files to modify: - `overlays/westside-ai-assistant/prod/deployment-patch.yaml` line 16 — change `claude-sonnet-4-20250514` to `claude-sonnet-4-6-20250514` Files NOT to touch: - Application code — model is read from env var - Base deployment — override stays in the overlay ### Acceptance Criteria - [ ] ANTHROPIC_MODEL env var set to `claude-sonnet-4-6-20250514` in kustomize overlay - [ ] Nemo responds correctly after ArgoCD sync - [ ] Latency acceptable (under 5s for typical questions) ### Test Expectations - [ ] Manual: ask Nemo a question, verify response quality - [ ] Manual: verify deployed env var matches ### Constraints - Persist in kustomize overlay (not kubectl set env) - Monitor Anthropic API costs after upgrade ### Checklist - [ ] PR opened - [ ] Tests pass - [ ] No unrelated changes ### Related - `project-westside-basketball` - westside-ai-assistant #27 (db_query tool — benefits from better model)
forgejo_admin commented 2026-03-31 01:11:58 +00:00
Author
Owner
Copy link

Scope Review: NEEDS_REFINEMENT

Review note: review-730-2026-03-30

The ticket's premise is incorrect: the current model is already claude-sonnet-4-20250514 (Sonnet 4), not claude-haiku-4-5-20251001 (Haiku 4.5) as stated. This is a Sonnet 4 → Sonnet 4.6 upgrade, not a Haiku → Sonnet upgrade.

Issues found:

  • [BODY] Current model value is wrong — actual is claude-sonnet-4-20250514 in pal-e-deployments/overlays/westside-ai-assistant/prod/deployment-patch.yaml line 16
  • [BODY] No file path specified — the change is in pal-e-deployments, not this repo
  • [BODY] Repo placement mismatch — issue filed on westside-ai-assistant but change lives in pal-e-deployments
  • [LABEL] story:WS-S5 does not match this work (WS-S5 = CNPG database / registration stats)
  • [SCOPE] Missing arch note: arch-westside-ai-assistant
  • [SCOPE] Missing user story for AI assistant model/quality management
## Scope Review: NEEDS_REFINEMENT Review note: `review-730-2026-03-30` The ticket's premise is incorrect: the current model is already `claude-sonnet-4-20250514` (Sonnet 4), not `claude-haiku-4-5-20251001` (Haiku 4.5) as stated. This is a Sonnet 4 → Sonnet 4.6 upgrade, not a Haiku → Sonnet upgrade. ### Issues found: - **[BODY]** Current model value is wrong — actual is `claude-sonnet-4-20250514` in `pal-e-deployments/overlays/westside-ai-assistant/prod/deployment-patch.yaml` line 16 - **[BODY]** No file path specified — the change is in `pal-e-deployments`, not this repo - **[BODY]** Repo placement mismatch — issue filed on westside-ai-assistant but change lives in pal-e-deployments - **[LABEL]** story:WS-S5 does not match this work (WS-S5 = CNPG database / registration stats) - **[SCOPE]** Missing arch note: arch-westside-ai-assistant - **[SCOPE]** Missing user story for AI assistant model/quality management
forgejo_admin changed title from Upgrade Nemo model from Haiku 4.5 to Sonnet 4.6 to Upgrade Nemo model from Sonnet 4 to Sonnet 4.6 2026-03-31 01:12:28 +00:00
forgejo_admin commented 2026-03-31 01:12:33 +00:00
Author
Owner
Copy link

Scope refinement (review-730-2026-03-30):

  1. Fixed model reference: was claude-haiku-4-5-20251001, actually claude-sonnet-4-20250514. Upgrade is Sonnet 4 → Sonnet 4.6.
  2. Fixed repo: change is in pal-e-deployments, not westside-ai-assistant.
  3. Added explicit file path: overlays/westside-ai-assistant/prod/deployment-patch.yaml line 16.
  4. Kept story:WS-S5 — this upgrade directly serves Marcus's ability to query dashboard/program data. Creating a new story for model management is overhead for a one-line change.
  5. Deferred arch note creation — arch-westside-ai-assistant is tracked as discovered scope, not a blocker for this ticket.
**Scope refinement (review-730-2026-03-30):** 1. Fixed model reference: was `claude-haiku-4-5-20251001`, actually `claude-sonnet-4-20250514`. Upgrade is Sonnet 4 → Sonnet 4.6. 2. Fixed repo: change is in `pal-e-deployments`, not `westside-ai-assistant`. 3. Added explicit file path: `overlays/westside-ai-assistant/prod/deployment-patch.yaml` line 16. 4. Kept `story:WS-S5` — this upgrade directly serves Marcus's ability to query dashboard/program data. Creating a new story for model management is overhead for a one-line change. 5. Deferred arch note creation — `arch-westside-ai-assistant` is tracked as discovered scope, not a blocker for this ticket.
forgejo_admin commented 2026-03-31 01:15:01 +00:00
Author
Owner
Copy link

Scope Review: APPROVED

Review note: review-730-2026-03-30-r2

Re-review after refinement. All blocking issues from the first review (review-730-2026-03-30) have been addressed:

  • Model reference corrected (Sonnet 4, not Haiku)
  • Repo corrected to pal-e-deployments
  • File path specified with line number, verified against codebase
  • story:WS-S5 justified (serves Marcus's data query capability)

Non-blocking items:

  • [LABEL] Board item #730 title still says "from Haiku 4.5" -- needs sync to "from Sonnet 4"
  • [SCOPE] Architecture note arch-westside-ai-assistant does not exist -- track as discovered scope

Ticket is ready for dispatch.

## Scope Review: APPROVED Review note: `review-730-2026-03-30-r2` Re-review after refinement. All blocking issues from the first review (review-730-2026-03-30) have been addressed: - Model reference corrected (Sonnet 4, not Haiku) - Repo corrected to pal-e-deployments - File path specified with line number, verified against codebase - story:WS-S5 justified (serves Marcus's data query capability) **Non-blocking items:** - **[LABEL]** Board item #730 title still says "from Haiku 4.5" -- needs sync to "from Sonnet 4" - **[SCOPE]** Architecture note `arch-westside-ai-assistant` does not exist -- track as discovered scope Ticket is ready for dispatch.
Sign in to join this conversation.
No Branch/Tag specified
Branches Tags
main
No results found.
Labels
Clear labels
No items
No labels
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
forgejo_admin/westside-ai-assistant#28
No description provided.