Skip to content

SLCP — Scotts Lawn Care Program

Live · Function: Consumer / Digital · R&D · Where it runs: Cloud Run + Cloud SQL + Firebase Hosting

The internal recommendation manager behind the Scotts Lawn Care Program — where R&D, product, and digital teams configure, preview, and publish the lawn-care plans Scotts customers receive.

Problem

The Scotts Lawn Care Program (SLCP) is the structured set of personalized recommendations Scotts delivers to consumers across the year — what to apply, when, in what conditions. The program needs continuous tuning by R&D and product teams as products change, seasons shift, and feedback comes in.

Before SLCP, that tuning happened across spreadsheets and ad-hoc tooling — slow, error-prone, and hard to audit.

What it does

  • Configure recommendations — R&D and product teams manage the recommendation corpus through a Vue 3 web app, with full preview against representative user profiles before publishing.
  • Chat-driven exploration — embedded chat-server interface lets internal users ask natural-language questions against the recommendation knowledge base (e.g. "what would we recommend for a northern fescue lawn in early spring?").
  • MCP-exposed tools — exposes recommendation lookups as MCP server endpoints so other internal agents and Bloom skills can call them.
  • Auditable changes — every recommendation update is versioned and traceable back to the change author.

Who owns it

  • Business owner: R&D + Consumer Digital
  • CoE AI Solution Lead: [TBD]
  • Tech stack: Vue 3 frontend (Firebase Hosting), FastAPI/Python backend (Cloud Run, private), Cloud SQL Postgres, Gemini-backed chat-server, MCP server

Where it runs

  • Frontend: slcp-app/web-app — Vue 3 + Vite, hosted on Firebase
  • Backend services: slcp-app/backend-services — backend API, chat-server, MCP server, all on Cloud Run
  • Architecture: Frontend → BFF (Cloud Function apiProxy with JWT validation) → private Cloud Run backend → Cloud SQL Postgres (via PSC)
  • GCP project: mkt-dev-0-0

Data it touches

Data source Classification Access path
Recommendation corpus Internal Cloud SQL Postgres, BFF-gated read/write
Product catalog Internal Synced from source systems
Internal user identity Internal JWT from BFF
LLM provider (Gemini) for chat — No PII in prompts; queries are about the recommendation corpus, not consumer accounts

Governance status

  • AI Governance Committee reviewed: Yes — internal tool, no consumer-facing surface, no PII in LLM prompts
  • Human-in-the-loop required for: All recommendation publishes require a reviewer
  • Vendor / model risk noted: Gemini usage scoped to internal corpus Q&A

Patterns reused

  • BFF → private Cloud Run pattern (reused by other internal CoE-built tools, including the AI Hub itself via IAP)
  • MCP-server-exposing-internal-tools pattern (foundation for agentic workflows in Bloom and future agents)

What's next

  • Deeper integration with MyLawn Brain so SLCP becomes a managed surface on top of the same rule engine
  • Expanding chat coverage beyond recommendation lookup to scenario simulation
  • Champion-network onboarding (FY27) so R&D team members can self-serve more changes

Changelog

  • 2026-05-21 — Initial entry on the hub.