LabCaddy
LabCaddy's users — researchers, lab managers, procurement teams — were searching for scientific equipment and consumables using keyword-based filtering that couldn't understand intent, context, or the relationships between products.
A researcher looking for 'something to measure pH in high-temperature samples' would get hundreds of irrelevant results because the search system matched keywords, not meaning. Domain-specific terminology made this worse — the same product might be described differently across suppliers.
The client wanted to move from keyword search to conversational product discovery — letting users describe what they need in natural language and get intelligent, contextually relevant recommendations.
- Designed a custom AI conversation flow for scientific use cases
- Built an AI-powered search layer for science-related products
- Enabled chatbot-driven search and product discovery
- Aligned the system around domain-specific language and workflows
We started by mapping the domain — understanding how researchers actually search for products, what language they use, and where existing search systems were failing them.
The conversational AI was trained to understand scientific context: it could differentiate between 'centrifuge for protein purification' and 'centrifuge for blood samples' and recommend appropriately different products.
The search layer combined semantic understanding with structured product metadata, so results were both contextually relevant and filtered by practical constraints like compatibility, certification, and supplier availability.
We designed the conversation flow to feel like talking to a knowledgeable lab assistant — asking clarifying questions when intent was ambiguous, rather than just returning best-guess results.
Conversational AI search deployed across the full product catalog — product discovery conversion improved measurably over keyword-only filtering.
The conversational search was deployed across the full product catalog, covering thousands of scientific products across multiple categories and suppliers.
Users who engaged with the conversational interface found relevant products faster and with fewer search iterations compared to the traditional keyword-based approach.
The domain-specific language model handled scientific terminology and contextual nuance that generic search systems consistently missed.
The system became a key differentiator in the platform's value proposition, with users specifically citing the search experience as a reason for continued use.