Group 1
Health & Wellness
KIND · RXBAR · Trü Frü · Nature's Bakery
GLP-1 tailwind. Competing for shelf space in a category retailers haven't yet decided how to organize.
Mars Snacking × Lyzr · Commercial Intelligence Blueprint
Overview
The Leader's intelligence advantage.
Mars Snacking spans sweet, salty, health, and premium — the broadest portfolio in the industry. That breadth is a durable advantage. It is also the most complex commercial intelligence challenge any snacking company faces.
$36BMars Snacking Revenue
9Billion-Dollar Brands
145+Global Markets
Prepared June 2026Sources SEC filings · Mars / Kellanova disclosures · Industry research
Section 01 · Portfolio
Three Portfolios. Three Market Realities.
Post-Kellanova, Mars Snacking operates across three distinct strategic contexts — each with different dynamics, different competitive sets, and a shared challenge.
Group 2
Core Brands
Snickers · M&M's · Pringles · Cheez-It · Twix · Skittles · Pop-Tarts
Nine billion-dollar brands. The value lever: getting more from every dollar of trade spend.
Group 3
Next-Gen Snacking
Hotel Chocolat · Premium Lines · Innovation Pipeline
Premiumization and new formats. Where Mars places its next portfolio bets.
Meanwhile, the Industry is Investing
Mondelez
Revenue Growth Management as stated #1 strategy; restructured around faster commercial decisions.
Hershey
"One Hershey" integrated commercial model; AI-powered shelf compliance and demand forecasting.
Competitors are building commercial intelligence capabilities. Mars faces a more complex version of this challenge — broader portfolio, wider commodity exposure, more categories to synthesize. That complexity is also the advantage.
+ Signal · Portfolio & Competitive Detail
PORTFOLIO DYNAMICS
Health & Wellness: GLP-1 users reduce snack consumption 40–60% but protein-rich snacking is up among users. 12.4% of US adults now on GLP-1 drugs (Gallup, 2025). Category CAGR 8%+.
Core Brands: Cocoa prices ~$5,000–6,000/tonne, 2–3x historical norms (J.P. Morgan). Industry-wide: 71% of US promotions fail to break even (Eversight). Trade spend typically 20%+ of gross revenue (McKinsey).
Next-Gen: Premium chocolate sales growing among GLP-1 users — "less is more" mindset (ConfectioneryNews, Mar 2026). Hotel Chocolat: 130+ stores, entering US market. Mars $42M R&D hub in Chicago.
COMPETITIVE DETAIL
Mondelez: Revenue Growth Management core strategy; combined CFO/COO role for commercial accountability; 2026 guidance 0–2% organic growth. (Sources: Mondelez DEF 14A FY2026; 8-K Q4 2025)
Hershey: "One Hershey" cross-category commercial model; AI demand forecasting for seasonal peaks; image recognition for shelf compliance; S/4HANA rollout. (Sources: Hershey Investor Day 8-K 2026; Consumer Goods Technology)
Section 01 · Portfolio
Three different market realities. One shared need: intelligence that drives revenue at the retailer interface and protects margin across the commodity landscape.
Section 02 · Intelligence
Two Nerve Centers. One Platform.
One drives revenue — account intelligence, promotion optimization, distribution expansion. The other protects margin — commodity intelligence, reformulation timing, price-pack precision. Together, they address both sides of the P&L.
REVENUE DRIVER
Retail Account Intelligence
Powers account managers at the retailer interface
Mars's best commercial intelligence reaches the top 5 retailers but not the top 50. Account managers at mid-tier and regional retailers prepare for category reviews with templated decks instead of synthesized, scenario-ready narratives.
Data inputs
- Syndicated POS data (NielsenIQ / Circana)
- Internal shipment & depletion data
- Trade promotion spend & ROI actuals
- Retailer-specific portal data (where available)
- Promotional calendars & competitive intelligence
Application by portfolio segment
Health & WellnessCategory builder — negotiates for prime shelf space and new SKU authorization
Core BrandsYield optimizer — sharpens trade promotion ROI and tightens retail execution
Next-Gen SnackingLaunch navigator — pre-tests channel entry and cannibalization before commitment
+ Signal · The Account Manager Challenge
A Mars account manager preparing for a Safeway category review today pulls from 5–6 disconnected sources: NielsenIQ/Circana syndicated data (2–4 week lag), internal shipment/depletion data, trade spend actuals, retailer-specific portal data (if available), promotional calendars, and competitive intelligence.
This synthesis is manual. For top-tier accounts (Walmart, Costco, Kroger), dedicated analyst teams support the process. For the hundreds of mid-tier and regional retailers — Safeway divisions, HEB, Publix, Meijer, convenience chains — account managers operate with significantly less analytical support.
The result: Mars's best insights reach the top 5 retailers but not the top 50. An agentic system democratizes insight quality across the full account portfolio — every account manager gets the equivalent of a dedicated analyst.
Post-Kellanova, the account manager may now have to cover chocolate, salty, AND health & wellness in a single meeting — synthesizing three category conversations simultaneously. No competitor faces this level of cross-category complexity.
MARGIN PROTECTOR
Input Cost & Product Intelligence
Powers product managers and commercial planning
Post-Kellanova, Mars has the widest commodity exposure in the snacking industry. No other company simultaneously manages cocoa, sugar, wheat, potato, dairy, and palm oil at this scale. Today, commodity intelligence sits in procurement, pricing in commercial planning, reformulation in R&D — rarely unified.
Data inputs
- Commodity price feeds (cocoa, sugar, wheat, dairy, palm oil)
- Regulatory signals (EUDR, HFSS, sugar taxes, packaging)
- Competitive pricing & reformulation moves
- Consumer elasticity & preference data
- Internal cost structures & hedging positions
The commodity landscape
CommodityProducts ExposedCurrent Signal
CocoaSnickers, M&M's, Twix, Dove, Hotel ChocolatPrices 2–3× historical norms
SugarSkittles, confectionery, baked snacksHFSS regulations spreading
Wheat / PotatoPringles, Pop-Tarts, Cheez-ItTariff exposure
DairyChocolate & baked categoriesScope 3 pressure
Palm OilReformulation ingredientEUDR Dec 2026
PackagingAll productsRecyclability mandates
+ Signal · The Leader's Commodity Challenge
Mondelez is primarily exposed to cocoa and wheat. Hershey is primarily cocoa and sugar. PepsiCo/Frito-Lay is primarily potato, corn, and oil. Mars post-Kellanova spans ALL of these simultaneously.
This creates a unique cross-commodity intelligence need:
· When cocoa moves +15% but wheat is stable, should promotional emphasis shift from chocolate to salty snacks?
· If EUDR compliance adds cost to cocoa sourcing, what's the reformulation trade-off vs. price-pack architecture adjustment?
· If GLP-1 is suppressing chocolate volumes but hedging positions are locked at high prices, what's the optimal response — absorb margin compression or accelerate premiumization?
These questions span procurement, commercial planning, and R&D. They connect raw material movements to promotional strategy, reformulation decisions, and price-pack architecture across the entire portfolio. Today the analysis happens across three functions, over weeks. The Margin Nerve Center runs it as one workflow.
Cocoa data: J.P. Morgan forecasts structurally higher at $6,000/tonne. Wholesale prices down ~70% from late 2024 peak but retail chocolate prices still ~14.4% elevated (Datasembly, early 2026). EUDR deforestation-free certification required Dec 2026 for large operators.
Sources: J.P. Morgan Research; ConfectioneryNews (Feb/Mar 2026); Datasembly; Just2Trade (Apr 2026); Sierra Club (Feb 2026).
One Agentic Platform
Two nerve centers, one platform, both sides of the P&L. The revenue nerve center makes every account manager as effective as the best. The margin nerve center turns commodity volatility into a planning advantage. Here's how both build — stage by stage.
Continue ↓Bridge · From Sequence to Evidence
The journey shows the sequence. What follows shows what's already built. The Mars configuration assembles from capabilities Lyzr runs in production today.
Section 03 · Journey
Your People Know the Business. Now They Build the Intelligence That Runs It.
Four stages. Each builds the data, trust, and capability the next stage requires. Lyzr's role shrinks as Mars's capability grows.
Efficiency
Intelligence
Advantage
Compounding
Core Brands
Category Review Synthesizer
Synthesizes 5+ data sources into a meeting-ready retailer narrative.
→ 2–3 days → 2–3 hours
+ Signal · Inside the Category Review
The account manager preparing for a Kroger confectionery review today synthesizes from 5–6 disconnected sources: NielsenIQ/Circana syndicated data (2–4 week lag), internal shipment/depletion data, trade spend actuals and ROI, Kroger 84.51° portal data, promotional calendars, and competitive shelf intelligence.
This synthesis is manual and quality varies — top-5 accounts get dedicated analyst support; mid-tier and regional accounts get templated decks. An agentic system democratizes insight quality across the full account portfolio.
The agent adapts tone and emphasis based on retailer priorities: Walmart values EDLP compliance and inventory turns. Kroger prioritizes loyalty-driven promotional effectiveness. Target focuses on trend alignment and demographic fit.
Sources: NielsenIQ; Circana; 84.51°; account-team interviews (2026).
Enablers Across the Journey
Consulting
Identifies and prioritizes use cases — top-down and Architect bottom-up.
FDE
Builds enterprise-grade agents on the Agentic OS.
Training
Transfers build capability — Architect first, then full Agentic OS.
All three draw from a shared knowledge base — proven agent patterns, reference architectures, and deployment playbooks from Lyzr's enterprise engagements. Consulting brings the patterns. FDEs build with them. Training transfers them to Mars teams.
Stage 01
Stage 02
Stage 03
Stage 04
Consulting Framework: How We Identify & Prioritize›
Lyzr's consulting team applies a structured scoring and prioritization framework to every engagement. Below are illustrative examples from existing enterprise deployments.
Where Each Use Case Lands Across the Three Dimensions
Operational impact, technical feasibility, and platform leverage — scored for prioritization.
Sequencing the Portfolio for Compounding Value
Use cases mapped into a deployment sequence — Deploy Now, Invest & Build, Quick Wins, Investigate
These frameworks are applied jointly with Mars leadership during the discovery sprint — scored with internal data to produce a prioritization specific to Mars Snacking's three portfolio realities.
JOURNEY · FROM EFFICIENCY TO COMPOUNDING
From Efficiency to Intelligence to Advantage to Compounding. Lyzr's role shrinks at every stage. Mars's capability grows. The destination isn't a vendor relationship — it's a Mars-owned commercial intelligence platform that no competitor can replicate.
Section 04 · Evidence
What's Built. What We Build With You.
Revenue and Margin Nerve Centers assemble from capabilities Lyzr runs in production — agent orchestration, retrieval, human-in-the-loop workflows, control plane, Architect authoring. The Mars configuration is built during the pilot — measured in weeks, not quarters.
The platform, illustrated
Analytics control tower
Shown: business outcome forecast, next agent action, quantified performance. In the Mars pilot: net revenue by category, trade promotion ROI, next-best action across the account portfolio.
Demand forecasting agent
Shown: agent architecture for quantitative scenario modelling. In the Mars pilot: category-level demand forecasting incorporating GLP-1 impact, weather, retailer promotional calendars, and syndicated panel data.
Order management with human approvals
Shown: human-in-the-loop pattern for high-stakes decisions. In the Mars pilot: trade promotion commitments, category review approvals, pricing changes — agents recommend, humans commit.
Structured product catalog at scale
Shown: structured operational data handled at meaningful scale. In the Mars pilot: SKU-level intelligence across retailer, category, and geography — the data foundation both Nerve Centers require.
Commercial intelligence for consumer brands
The Lyzr Commerce OS trend sensing capability. Signal detection across markets, audiences, and platforms. Configured here for consumer product categories — the same architecture that powers brand and category intelligence for Mars.
Trend signal to strategyMulti-market audience insightScheduled monitoring
In the Mars pilot: signal detection across confectionery, salty, and health & wellness. Emerging trend intelligence for brand teams. Consumer shift detection ahead of syndicated panel data.
One platform. Two views. The Nerve Centers Mars builds run on the same foundation.
We assemble. We do not research.
Section 05 · Architecture
Nothing Replaced. Everything Connected. Production at Speed.
Gemini builds your AI foundation. Writer powers your content. Lyzr adds the enterprise controls both platforms need — governance, observability, and continuous refinement.
Built for the stack Mars already chose
Mars has deepened its partnership with Google on Gemini Enterprise, and with Writer on generative AI. Lyzr runs on both — extending, not replacing.
The Lyzr Agent Platform is designed to sit alongside enterprise foundation model commitments. Mars's existing infrastructure choices become the foundation. Lyzr adds the agent production layer, the governance, and the observability that both platforms need to move from pilots to production.
Source: Lyzr × Writer partnership documentation (lyzr.ai/lyzr-and-writer)
One Production Layer. Every Agent Platform.
Lyzr + Gemini Enterprise▼
Build agents with Google. Ship them to production using Lyzr.
Google Cloud builds the agent runtime. Lyzr is the production layer that turns Mars's Gemini Enterprise investment into governed, observable, continuously improving enterprise software.
Capability
Gemini / Vertex AI handles
Lyzr adds on top
Model & Runtime
Vertex AI, Gemini, Model Garden, Model Armor, Vertex Responsible AI
Routes Gemini only to the work that needs it — Six Sigma architecture sends routine tasks to smaller, cheaper models to lower inference TCO without sacrificing quality
Agent Building
Vertex AI Agent Builder and ADK — code-first tools for engineers
SuperFlow canvas for no-code agent design; Architect enables natural-language agent building on top of ADK and any framework
Data & Grounding
BigQuery, AlloyDB, Firestore, Vertex RAG Engine
Agentic RAG + GraphRAG across heterogeneous sources; natural-language to SQL on BigQuery; Cognis persistent memory at 92.4% LongMemEval accuracy
Deployment & CI/CD
GKE, Cloud Run, Compute Engine — compute infrastructure for running agents at scale
Real agent CI/CD: non-prod → pre-prod → prod promotion with approvals, versioned releases, instant per-release rollback — today, not on a roadmap
Safety & Quality
Model Armor and Vertex Responsible AI on every model call — policy enforcement at the inference layer
ASIM simulation engine — test agents against thousands of real-world scenarios before they touch production
Cross-Vendor Governance
Observability and IAM within GCP — best-in-class inside the Google Cloud environment
One control plane across Vertex AI Agent Builder, Agentspace, Agentforce, ServiceNow, Databricks, and custom agents — one entitlement model, one audit trail
Continuous Improvement
Logging, monitoring, and cost attribution via Cloud Logging and BigQuery
Improvement Engine — RL-based drift detection that catches when agent behavior shifts from intent and corrects it automatically
Nothing in Google Cloud gets replaced. Vertex AI stays primary. Mars data never leaves the GCP VPC. Lyzr adds the production layer that sits above — and extends the same governance to every other agent system Mars runs.
For additional informationLyzr + Writer▼
Writer runs your content agents. Lyzr connects them to the rest of Mars's enterprise.
Extending the Writer investment beyond content into full agent orchestration.
Capability
Writer AI HQ handles
Lyzr adds on top
Agent Building
Visual Agent Builder, 100+ templates, AI Guardrails
Wraps Writer agents into cross-vendor orchestration without touching what's built inside Writer
Knowledge & Memory
Knowledge Graph with multi-hop reasoning
Cognis — persistent memory that travels across sessions, tools, and vendors
Orchestration
MCP gateway for external tool calls
SuperFlow — wire Writer agents alongside Gemini, Salesforce, and custom agents in one workflow
Deployment
Activate and supervise from AI HQ
CI/CD with environment promotion and instant rollback — today
Risk & Quality
AI Guardrails at content level
ASIM simulation — test agents against real scenarios before production
Governance
Role-based access within AI HQ
One entitlement model and audit trail across Writer AND every other vendor
Writer agents stay in Writer. Palmyra stays primary. Lyzr connects Writer's agent layer to the rest of Mars's enterprise.
For additional informationTwo agent platforms Mars already uses. One production layer that governs both. Zero rip-and-replace.
SOC 2HIPAAGDPRISO 27001VPC-NativeSSO/RBAC
Section 06 · Trust
Enterprise-Ready.
Backed by the strategic partner Mars already works with. Built to enterprise standards.
Backed by Accenture. Trusted by Deloitte and KPMG.
The strategic investors and system integrators Mars works with have already validated Lyzr — with capital, with delivery investment, and by building their own consulting IP on the platform.
Three investments in nine months
Accenture Ventures has invested in Lyzr three times within nine months. Unprecedented in Accenture's history.
$65M pipeline in 90 days
Accenture Song added $65M to their pipeline in the first 90 days of deployment on Lyzr. Every new Song experience is now built on the platform.
Big Four building their IP on Lyzr
Deloitte and KPMG are building their proprietary consulting IP on the Lyzr platform. Not evaluating — building.
Mars's existing consulting partners have already invested time, money, and reputation in Lyzr. That validation is a starting point, not a claim.
Recognized by leading analysts. Built to enterprise standards.
Independent analyst recognition and enterprise-grade certifications from day one.
Source: Lyzr analyst recognition and compliance summary, 2026
Enterprise-grade infrastructure. Enterprise-grade partners. Ready for Mars from day one.
Bridge · From Trust to Action
A platform earns the right to scale by proving value on one category, one retailer, one quarter — then compounds.
Section 07 · Next Steps
Let's Start.
Confirmed protocol before wider sharing.
- 1Suneet completes a self-review of the presentation.
- 230-minute walkthrough with Suneet before any wider sharing.
- 3Wider distribution only after the walkthrough.
Mars built the world's broadest snacking portfolio. Lyzr builds the intelligence that turns that breadth into a revenue and margin advantage no competitor can replicate.