A leading global consumer appliances company came to Armakuni for multi-program GenAI transformation: competitive pricing intelligence, IoT telemetry analytics, AI-powered anomaly detection, synthetic consumer research, recipe assistant, and dynamic content generation.
The challenge
A GenAI discovery assessment commissioned in June 2025 identified two priority capabilities and produced a validated roadmap. Armakuni then executed against that roadmap across seven SOWs spanning competitive intelligence, IoT analytics, anomaly detection, synthetic consumer research, recipe automation, and dynamic image generation. Each SOW built on shared AWS infrastructure and Amazon Bedrock capabilities established in earlier phases.
Scale: Global consumer appliances brand operating across North America and international markets. IoT fleet generating 70,000+ telemetry records per minute. Products include robotic vacuums, smart grills, and connected kitchen appliances sold across hundreds of retailers.
What we built
Armakuni delivered multi-program GenAI transformation: competitive pricing intelligence, IoT telemetry analytics, AI-powered anomaly detection, synthetic consumer research, recipe assistant, and dynamic content generation. The engagement ran as multiple fixed-scope and time-and-material SOWs: GenAI Readiness Assessment, IoT Data Insights POC, IoT Anomaly Detection POC, Product Scraping & Analytics (Mexico Pilot + North American P2P), Synthetic Persona Intelligence Platform POC, Recipe Assistant POV, Dynamic Image Generation Tool. The work was AWS funded.
The outcomes
6 measurable outcomes shipped across the engagement. The ones that moved the business the most:
Competitive pricing intelligence live in production across 160 retailers and 1,150+ SKUs in three countries. Automated buy box tracking, 1P/3P seller identification, enforcement screenshot capture, and a six-times-daily refresh for major retailers, capabilities that did not exist for Mexico at engagement start.
IoT telemetry reaches non-technical product teams through natural language for the first time. Analysts ask questions about fleet performance without SQL, receive automated daily and weekly reports, and act on anomaly alerts that previously required manual investigation.
Real-time anomaly detection now runs against the live device fleet. SageMaker models classify anomalies by severity, predictive maintenance schedules come from usage patterns, and SNS alerts route issues to the right team before customers feel the consequences.
Consumer Insights now runs synthetic exploratory studies in-house without an external vendor. The platform processes screener documents, generates grounded synthetic personas, and produces qualitative insights across six CI reporting dimensions in days instead of weeks.
Social video to appliance-optimized cooking instructions in one automated workflow. The proof of value demonstrated extraction accuracy from TikTok and Instagram URLs with dietary customization and appliance-specific adaptation in a full end-to-end prototype.
Geo-expansion content pipeline removes the manual design bottleneck on market entry. The pipeline translates promotional banners to any target language and currency at resolution levels suitable for both web and print distribution, unblocking the company's pace of market entry.
Built on AWS
Delivered under our AWS AI Services, Migration and Modernization competencies.
What's next
Path-to-production scaling of the Product Scraping platform through Q3-Q4 2026 including self-service retailer onboarding and comprehensive monitoring dashboards. Production deployment of IoT analytics and anomaly detection capabilities. Expansion of synthetic persona platform to additional research domains.