Decision AI for supply chains, built in Zürich
Taming the Bullwhip Effect in FMCG & Retail Supply Chains.
A small change at the retail shelf becomes a crisis at the supplier. Kanvaz finds where the signal breaks between tiers — and fixes the cause before reactive decisions amplify the waste.
01 — The problem
Every tier amplifies the noise. Excess stock on one end, stockouts on the other.
A small swing at the retail shelf grows to five times the variance by the time orders reach raw-material suppliers. Each tier adds safety stock, batching and guesswork.
Stage 04
Raw suppliers
5.16×
90th-percentile manufacturer
Stage 03
Tier-1 assembly
3.04×
Mean manufacturer
Stage 02
Distribution
1.28×
Median amplification
Stage 01
Retail POS
1.0×
Demand baseline
Proof of impact
12.5–25%
of total supply chain cost can be removed by eliminating bullwhip distortion.
Lee, Padmanabhan & Whang — MIT Sloan Mgmt Review, 1997
Up to 4×
order variance amplification measured from retail to upstream suppliers in consumer goods.
Lee et al. — P&G Pampers case, Management Science, 1997
20–50%
lower inventory reported when planning runs on shared, trusted demand signals.
McKinsey — AI-driven supply chain planning, 2021
02 — The platform
Middleware between your systems and planning. Three layers.
Trust the signal
GlossDB
A data middleware that harmonizes and checks every load before it reaches planning. Your systems stay.
Find the cause
Causal AI
Separates true demand from distortion and tests interventions before you change a parameter.
Act with control
Human–agent workflows
Agents surface exceptions daily. Your planners approve what matters.
03 — How to start
Start with a free 1–2 week proof
of concept.
1–2 weeks · Free
Proof of concept
Find bullwhip drivers on a sample of your own data. No obligation to continue.
6–8 weeks · Fixed fee
Pilot
Validate one market and one data domain against agreed success criteria.
3 months+ · Milestones
Roll-out
Extend to more markets and teams. Pay against accepted milestones.
Ongoing · Subscription
Maintain
Keep models, definitions and workflows accurate as your business changes.
Read-only · your planning systems stay in place · go / no-go at the end of each phase
Contracts and guarantees by phase

