BFC sought to streamline how they handled incoming customer inquiries. For each new request, the team would cross-reference multiple data sources to find comparable past pieces. Surface area, clay quantity, glaze, kiln time, and labor all had to be worked out from scratch, regardless of how similar it was to something priced earlier.
A single multi-item request could take hours to process. Beyond the quoting stage, the same information had to be re-entered manually into their financial system once an order was confirmed, meaning the effort effectively doubled. As order volumes grew, so did the workload.
BFC's vision was to build a searchable pricing knowledge base, tightly integrated with their quoting and financial workflows, so that increased demand would translate into more business.
Aivar partnered with BFC to design and deploy an AI-powered quoting assistant, built on AWS and driven by Anthropic's Claude on Amazon Bedrock. The platform automates the full pricing cycle, from the moment an inquiry arrives to the point a quotation is ready to send and an entry is created in the financial system.
When a new inquiry comes in, Anthropic's Claude, running on Amazon Bedrock, reads the product descriptions and images, searches BFC's entire historical catalog using both text and visual matching, and surfaces the most relevant past items along with a clear, plain-language explanation of why each was selected. Rather than returning a black-box suggestion, Claude walks the operations team through the reasoning, so every recommendation can be reviewed with confidence.
Pricing fields are pre-populated based on the matched items. The team reviews, adjusts where needed, and confirms. The outgoing quotation is then generated automatically, with no manual assembly required. When a purchase order is approved, a single upload triggers Claude to parse the document, resolve the customer record, and create a fully pre-populated draft in the accounting system, ready to review and send. Every item that passes through the platform is automatically re-indexed, meaning the catalog grows smarter with every order cycle.
The solution is built on AWS, with Amazon Bedrock serving as the AI layer. Anthropic's Claude acts as the core reasoning engine throughout the platform, responsible for understanding product context, ranking historical matches, generating pricing explanations in plain language, and processing purchase order documents end to end. Multimodal embeddings enable the system to match on both product images and written descriptions.
Security and reliability were treated as first-class requirements from the outset. The platform is built with encryption, access controls, audit logging, and resilience configurations that are appropriate for a production business environment and ready to scale as BFC grows.
The results were immediate and measurable across every stage of the workflow.
Aivar's engagement with BornFree Ceramics is an example of how we approach every AI build, understand the real operational challenge first, validate the technology against actual customer data before committing to a full build, and deliver something the team can genuinely use and trust from day one.
.svg)
.svg)
.svg)
.svg)
