Ideas
If something in your work felt slow, manual, or repetitive this week — tell us in 60 seconds. We read every one.
What happens after you submit
Every idea goes through the same lightweight triage. There's no "approval queue" and there's no black hole.
flowchart LR
A[You submit] --> B[Triage<br/>weekly]
B --> C{What's the right path?}
C -->|Enterprise-wide benefit| D[AI Solution Lead picks it up]
C -->|Valid idea, not now| E[Backlogged]
C -->|Do this today locally| F[1:1 sharing<br/>Gems or replicate workflow]
C -->|Foundation not ready| G[Parked with reason]
D --> H[Solution Library entry]
F --> HEvery idea gets a status visible on this page — even "parked, here's why" is a real answer.
What we're looking for
Good intake submissions sound like:
"I spend ~2 hrs every Friday reconciling X spreadsheet with Y export. The matching is 80% deterministic."
"Every new product launch, our team writes ~30 SEO descriptions by hand from spec sheets. Quality drifts late in the day."
"I want to chat with our consumer research archive — it lives across PDFs and SharePoint and we can never find what's there."
What's less useful:
- "AI for marketing" — too vague, no problem
- "Build me a chatbot" — solution-first, no problem
- "What can AI do for X?" — better as a topic suggestion for office hours than an intake
What's in the queue
Below is the active queue — one card per submitted idea, with the CoE's read on what it is, what's been built (if anything), and the triage path. The cards are hand-curated today; a Bloom triage agent will automate this once it's online.
Cutoff
We show submissions from 2026-05-12 onward — earlier entries on the intake board were migrated from prior tooling and aren't part of the active queue.
Production Planning Tool
Problem. Growing-media production sites face daily shortages and shifting customer priorities. Deciding what to run today to capture the most revenue required pulling fragmented data out of SAP and reasoning across business priorities, line capabilities (BPM rates, SKU-line compatibility), and customer-demand signals — 20–30 minutes of focused work every morning.
Proposed solution (already built). Andrew used Gemini API + Workspace to (1) pull SAP data into one place via generated code, (2) implement a priority-ranking algorithm aligned with the supply-chain/sales priority logic already in Sigma, and (3) wire it all to a Google Sheets button that produces the daily plan in 2–3 minutes. Surfaces both what to run and which customer-specific replenishment opportunities exist.
Tri-Engine AI Operations Framework
Problem. Brand managers face information bottlenecks — fragmented high-velocity data (vendor/agency emails, industry feeds, POS performance, cross-functional dependencies). Three symptoms: information fatigue (insights slip through), analysis friction (qualitative alignment vs. hard metrics), execution lag (cognitive overload converting raw data to prioritized daily action).
Proposed solution (already built). A "Tri-Engine" Workspace setup spanning three horizons: Macro-Strategic (weekly review with variance testing and branded-vs-private-label deconfliction), Micro-Tactical (24–72h work briefing with entity tokenization and margin-play detection), and Operational Governance (an "AI Chief of Staff" layer). Built on Gemini Workspace + API + NotebookLM.
Daily Business
Problem. Day-to-day data parsing in supply-chain spreadsheets is slow and manual. Wants help optimizing Excel workflows.
Proposed solution. Get a "Gemini Add-In" for Excel (colleagues at other companies have used this). Less of a specific use case, more of an enablement / tooling request.
Creative Brief writing
Problem. Writing creative briefs is slow, and finding the right language to guide creative teams/agencies is hard.
Proposed solution (already built). Uses Gemini API — feeds AI brand background + brief template + desired outcomes, lets the model help complete the brief. Personal workflow.
Blocked Stock Workflow
Problem. Riverside facility manages blocked stock and material rework via Google Sheets. The business logic (translating finished SKUs to raw mix yardage) is solid; the orchestration (task tracking, reminders, SAP data entry for 343/344 stock movements) is entirely manual. Creates bottlenecks reclaiming raw materials and ties up valuable yard space.
Proposed solution. PoC autonomous AI agent on the Google Cloud Agent framework. Starts with the structured 7-day SLA blocked-stock workflow — agent ingests data, authenticates, performs logical checks, orchestrates the SLA. Stretch goal: scale to predictive maintenance + Vorne OEE integration.
Streamlining Product Knowledge Across All Platforms
Problem. Consumer Services + R&D content is scattered across labels, SDS, PDP URLs, internal articles. Symptoms: inconsistent product knowledge across channels, slow product-launch audits (manual old-vs-new label comparisons), agent cognitive load on live interactions, and brand-voice/safety-voice disconnects.
Proposed solution (already built). Five custom Gemini GEMs covering: (1) "single source of truth" content synthesis from labels/SDS/PDP; (2) automated label-comparison delta reports for new product launches; (3) UX-optimized articles using Cognitive Load Theory and F-shape scanning; (4) brand-voice alignment GEM that translates technical compliance into unified voice; (5) NotebookLM-driven training-content acceleration.