/ Digital Transformation · Case 02 · Health Beverage · Singapore
Drinkizz
Smart Information System from zero — 3× production, 100% digital traceability, $30M Series A.
The brief.
Drinkizz needed an information system that could keep pace with the brand's growth — production, inventory, distributor operations, retail channels, and traceability all routed through one source of truth. As Digital Transformation Manager from 2019 through 2023, I designed the data architecture and built the Smart Information System from zero, on a no-code/low-code stack with AI augmentation, while operations scaled across two countries.
The IT infrastructure was built on a Knack no-code database, with Make handling automation, and ChatGPT layered in as multiple AI assistants for data modeling support, process automation, and knowledge exchange with clients and partners. PowerBI closed the loop with executive-grade dashboards. WooCommerce, Chatbase, Mailchimp, Simple Backup and Vietnam Blockchain International rounded out the stack — five-plus tools consolidated into a single, coherent operational system.
With Semantic Modeling and a Knowledge-Graph data system in place, we layered RAG (Retrieval-Augmented Generation) on top to enhance real-time decision-making — context-aware, cost-efficient insights that let LLMs and machine learning drive smarter automation, personalised experiences, and improved operational efficiency throughout the digital transformation journey.
What was on the table.
A scaling startup needed to manage complex production and inventory workflows without existing IT infrastructure — and do it across B2B and B2C in parallel.
How it shipped.
Designed data modeling and IT architecture aligned to the value chain — every domain analysed for its specific requirements before tooling decisions.
Built a No-Code/Low-Code Smart Information System on Knack + Make consolidating 5+ tools into one source of truth.
Integrated AI assistants (ChatGPT) and automation across production and retail, hitting 3× production with no added headcount.
Delivered 100% digital traceability ingredient-to-shelf, with real-time inventory visibility across both countries.
Layered RAG over a Semantic / Knowledge-Graph model for context-aware, real-time decision-making.
An information spine, not an ERP
The conceptual frame at Drinkizz was that a scaling consumer brand does not need an ERP — it needs an information spine. One source of truth that production, distribution, retail, and finance can all read from, with semantics owned by the business and tooling owned by IT.
The value chain is the model. Tooling follows the model, never the reverse.
Five tools consolidated into one spine is worth more than ten tools running in parallel.
AI assistants augment the spine; they do not replace its discipline.
Value-chain entities + semantic + Knowledge Graph
We mapped the value chain into key data domains — production, inventory, distributor, retail, traceability, finance — and modeled each as a small set of stable entities. Semantic Modeling unified definitions across domains; a Knowledge-Graph data system published relationships; RAG layered context-aware retrieval on top.
Six core domains, each with 3–5 canonical entities and named stewards.
Semantic layer enforced one definition of 'product', 'order', and 'outlet' across all tools.
Knowledge Graph captured the relationships traversal queries actually needed.
RAG retrieval served real-time decisions to operators on the floor and the road.
Knack + Make + ChatGPT + PowerBI on a No-Code spine
The physical implementation was deliberately no-code/low-code. Knack hosted the database, Make handled automation, ChatGPT ran as multiple AI assistants, PowerBI closed the executive loop. WooCommerce, Chatbase, Mailchimp, Simple Backup and Vietnam Blockchain International rounded out the stack — five-plus tools consolidated into one coherent operational system.
Database: Knack no-code database hosting the value-chain entities.
Automation: Make routes events between domains without bespoke ETL.
AI: ChatGPT assistants for modeling support, process automation, and partner knowledge exchange.
BI: PowerBI dashboards aligned to the semantic layer, not to ad-hoc SQL.
Edge: WooCommerce, Chatbase, Mailchimp, Simple Backup, VN Blockchain Intl.
“An MVP information system isn't smaller — it's just honest about which decisions it has to support today.”
Numbers, verified.
- Production capacity, no headcount
- 3×
- Digital traceability
- 100%
- Tools consolidated
- 5 → 1
- Series A valuation supported
- $30M
/ What we shipped
Value-chain data model + key data domains
No-Code/Low-Code Smart Information System (Knack + Make)
AI assistants (ChatGPT) for modeling, automation, knowledge exchange
Real-time inventory + 100% digital traceability
PowerBI executive dashboards
RAG layer over the semantic model
Tool consolidation: 5 → 1
What stuck.
The lesson from Drinkizz: a small, fast-moving company doesn't need an ERP — it needs an information spine. Once the spine was in place, every tool we added compounded. Without it, every tool would have fragmented the operation further and made the next hire harder, not easier.