Client
SyrenCloud
Mission

Democratise high-quality financial advice using generative AI

Pre-Engagement State

Python Flask API, OpenAI completions, MongoDB

SyrenCloud

SyrenCloud
Case Study

Executive Summary

SyrenCloud—a data-services arm of a €5 billion fast-moving consumer-goods (FMCG) group—struggled to meet its flagship KPI: OTIF-D (On-Time In-Full Delivery). Unreliable demand forecasts, siloed supplier updates, and spreadsheet-driven inventory targets left planners fighting fires. Each 1 % slip in OTIF-D cost an estimated €2.7 million in lost sales and penalties.

Steady Rabbit deployed a Micro-GCC Core-Flex squad that blended data-engineering muscle with on-demand supply-chain and Azure analytics SMEs. In only 24 weeks, the team:

  • Unified 24 data sources—ERP, MES, WMS, and supplier portals—into an Azure Synapse + ADLS lakehouse
  • Built XGBoost + Prophet hybrid forecasts that lifted MAPE accuracy 15 % (from 76 % → 87 %)
  • Launched Looker OTIF-D dashboards used by 300+ planners and suppliers, slashing manual reporting 90 %
  • Introduced a risk-heat-map & alerting engine that drove unit fill rate to 97 % and external-manufacturing adherence to 98 %
  • Delivered €12.4 million annual working-capital savings via inventory-optimization simulators
  • Passed ISO 27001 surveillance and SOC 2 Type I audits with zero major findings

Every sprint landed on time; no P0 incidents hit production in the first 120 days. SyrenCloud now markets the OTIF-D Suite as a paid add-on—projected €4.1 million ARR in year one.

syrencloud syrencloud website

Client Profile & Business Context

  • Client
    SyrenCloud

    Analytics subsidiary of a global FMCG conglomerate

  • Founded

    2019 (data-COE spinoff)

  • Mission

    Provide cloud analytics that unlock supply-chain agility

  • Regions Served

    14 factories • 62 third-party manufacturers • 120+ suppliers

  • Pre-Engagement Tech

    SAP ECC, Excel MRP, isolated PowerBI reports, Azure AD w/out RBAC

  • Strategic Trigger

    Board mandate: lift OTIF-D from 92 % → 97 % before FY close

Problem Statement / Key Challenges

Inaccurate Forecasts

Challenge

Sales forecasts MAPE 24 %; planners padded inventory → €38 M tied-up cash

Business Pain

Stock-outs & write-offs

Siloed Data

Challenge

ERP, MES, WMS, supplier EDI in separate clouds

Business Pain

No single OTIF-D version of truth

Reactionary Planning

Challenge

Excel macros produced weekly plans; changes took 48 h to ripple

Business Pain

Late PO releases, missed slots

Poor External-Manufacturing Visibility

Challenge

30 % output via TPMs; adherence 82 %

Business Pain

Penalties, expedited freight

No Proactive Alerts

Challenge

Issues surfaced after misses

Business Pain

OTIF-D stuck at 92 %, CFO scrutiny

Compressed Timeline

Challenge

6 months to meet KPI or cut SKU expansion budget

Business Pain

Strategic growth at risk

Our Approach

Micro-GCC Squad Structure

Layer
Composition
Mission
Core (7)
Squad Lead/PO, 2 Azure Data Engineers, ML Engineer, Looker BI Dev, DevOps/SRE, QA Automation
End-to-end lakehouse, ML forecasts, dashboards
Flex (2)
Supply-Chain SME (APICS CPIM), Azure Solution Architect (DP-203)
Spike tasks: MRP logic, Synapse performance
Buffer (1)
Shadow Data Engineer
PTO/attrition shield—funded by Steady Rabbit

Shift-Left Governance

  • Seven Plan-Left gates—Persona → Acceptance → Risk label → Arch sketch → Estimation → Capacity (via SteadCAST) → Test note.
  • SteadCAST dashboards posted Risk-High WIP, velocity drift at 9 a.m. daily.
  • 30-minute weekly steering with VP Supply-Chain & CIO—demo, KPIs, burn-rate.

Discovery Sprint 0 (Weeks 1-2)

  • Value-Stream Mapping — from demand signal → MRP → TPM shipment.
  • Architecture Blueprint — ADLS Gen2 lake ► Azure Databricks ETL ► Synapse DW ► Looker semantic layer.
  • North-Star KPIs — Forecast MAPE ≤ 15 %, Fill rate ≥ 97 %, TPM adherence ≥ 98 %, schedule compliance ≥ 95 %.

Outcome: backlog sized to 112 SP/sprint; go-live target Week 24.

Solution Delivered

Unified Lakehouse

  • AWS EKS cluster with Fargate isolation per environment.
  • RDS PostgreSQL (encryption-at-rest) + DynamoDB for session tokens.
  • AWS KMS envelope encryption; CloudTrail + GuardDuty for audit.
  • Terraform Cloud stored state & produced SOC 1 evidence.

Forecasting & Inventory Optimiser

  • React-Native reused 92 % code across iOS & Android, enabling parallel beta launch.
  • Next.js PWA delivered sub-1 s TTI on mobile web.
  • In-app WebRTC video (Daily.co) replaced free Zoom, ensuring PHI control and in-stream recording.

OTIF-D Dashboards & Heat Map

  • GPT-4 chatbot (LangChain) conducts empathetic intake, runs PHQ-9/GAD-7; auto-escalates high-risk scores.
  • Personalised content pipeline: vector search on pgvector suggests CBT modules; 63 % Tier-1 tickets auto-resolved.

Alerting & Collaboration

  • KYC verification, calendar sync (Google/Microsoft), and dynamic pricing in one flow.
  • Image-enhancement Lambda improves profile photos; average onboarding 12 days → 36 hours.

Third-Party Manufacturing Portal

  • Event listeners trigger push, email, WhatsApp reminders; no-show rate 40 % → 9 %.
  • Gamified progress bar encourages completion of therapy homework; day-14 churn 38 % → 15 %.

DevSecOps & Compliance

  • Grafana Loki for logs; Prometheus for latency, error budgets.
  • SOC 2 & HIPAA artefacts auto-exported for auditors saved HealExpert 120 staff hours.

Execution Journey

Phase
Timeline
Key Deliverables
Predictability
Sprint 0 (Weeks 1-2)
Discovery & architecture
Threat model, backlog, KPI baseline
100 % gate pass
Sprints 1-2
Cloud foundation, IaC, React-Native skeleton
p95 API latency 950 ms
Buffer unused
Sprints 3-4
AI triage bot POC, therapist portal MVP
First chatbot accuracy 81 %
Flex AI SME 24 h
Sprints 5-6
Video module, GraphQL BFF, compliance scans
Sonar bugs -91 %
Risk-High WIP < 15 %
Sprints 7-8
Push/WhatsApp reminders, load tests, blue-green drills
WebRTC QoS 99 %
Schedule slip 0 days
Sprint 9
Production launch, enterprise wellness PoC
Day-1 activation +23 pp
Budget variance +4 %
Sprint 10
Post-launch hardening, Series A deck support
Uptime 99.96 %
0 hot-fix Fridays

When the DevOps engineer faced a family emergency in Sprint 6, the Buffer dev stepped in within 3 hours, ensuring Terraform pipeline PRs merged on time—zero velocity impact.

Business Outcomes & Impact

User Activation 48 % → 71 % (+23 pp) within 30 days

No-Show Rate 40 % → 9 % (-31 pp) after automated reminders

Tier-1 inquiries auto-resolved 63 % via GPT-powered triage, cutting support FTEs by 1.5 headcount

Peak-hour p95 latency 950 ms → 420 ms (2.3× faster)

Therapist onboarding 12 days → 36 hours, enabling rapid supply scaling

Day-14 churn 38 % → 15 % through personalised CBT modules

HIPAA & ISO 27001 audit pass 2 months ahead; enterprise buyer contract signed USD 1.9 M

Secured USD 8 M Series A; investor memo cited “enterprise-grade tech backbone”

Predictability Premium ROI: Steady Rabbit’s 8 % blended-rate premium saved ~USD 930 k cost-of-delay by avoiding a projected four-week slip.

Why Steady Rabbit?

Core-Flex Micro-GCC Model

Right specialists appeared within 48 h; Buffer bench absorbed shocks.

Shift-Left Governance

Seven Plan-Left gates cut re-work 40 %, without heavyweight PMO.

SteadCAST Predictability

Real-time dashboards kept schedule variance ≤ 3 %.

AI & Compliance Expertise

Team included clinicians, CISSP architect, and LangChain SME—critical for HIPAA + GPT success.

Outcome-Linked Engagement

KPIs (activation, latency, audit) tied to squad incentives; no vanity metrics.

Transparent Partnership

Weekly steering, Slack war-room, shared burn charts—zero surprises.

Client Testimonial

Steady Rabbit

VP Supply-Chain Analytics

SyrenCloud

Steady Rabbit turned scattered data into an always-on command center. Forecasts improved overnight, stock-outs vanished, and we hit our OTIF-D goal a quarter early. Their Micro-GCC model is the definition of predictable delivery.