Chemicals software

// The problem

A coatings and adhesives maker sells in Canada, the US and Mexico, keeps safety data sheets and technical data sheets as files by language, and its customer service team spends hours emailing the current versions.

// What we would build

A customer portal on the company's brand with every product's current SDS and technical data sheet in English, French and Spanish, versions kept with their approval history, and an assistant that helps customers choose products from the technical data sheets. It runs on an open-source model in the company's own cloud account, so formulas and drafts of new sheets never leave it. For safety questions the assistant links the current SDS and never paraphrases it.

// What is in it

// Stack

• Next.js
• Llama or Mistral (open-source)
• vLLM
• PostgreSQL with pgvector
• ERP API
• AWS Canada

// The problem

A specialty chemicals plant runs batches from printed tickets, operators write weights by hand, QC results are in a separate spreadsheet, and tracing a customer complaint back to raw material lots takes days.

// What we would build

Electronic batch records on rugged Android scanners at each station: the operator scans each raw material lot, the scale weight is read, not typed, and steps follow the approved formulation. QC results are read from lab instruments against the batch, and each drum is labelled from the record. Formulations are restricted by role, with every view logged.

// What is in it

// Stack

• Next.js
• Kotlin for Android
• PostgreSQL
• Instrument and scale interfaces
• ERP API
• AWS Canada

// The problem

A plastics compounder has decades of formulation experiments in notebooks and spreadsheets, and its chemists start each customer request by asking the one person who remembers what was tried.

// What we would build

A custom model fine-tuned on the company's own formulation records and lab results, which suggests starting formulations for a target property and finds similar past experiments, with the records behind each suggestion. It runs on a GPU server the company owns, inside its own network, because the formulations are the business. Chemists decide what to test.

// What is in it

// Stack

• Llama (fine-tuned)
• PyTorch
• vLLM
• PostgreSQL with pgvector
• Next.js
• Lenovo GPU server

// The problem

A chemical distributor retypes UN numbers, classes and packing groups from the SDS and the order into two carriers' systems and its warehouse system, and a mismatch holds shipments at the dock.

// What we would build

An integration hub that takes the transport information from the approved SDS records and the order from the ERP, and passes it to the warehouse system and the carriers' booking APIs, so it is entered once. Shipping documents are prepared from this data for your trained dangerous goods staff to check and sign. Each connector is built on the system's published API; we check yours before we quote.

// What is in it

// Stack

• Node.js
• PostgreSQL
• OpenAPI
• ERP APIs
• Carrier booking APIs
• AWS Canada

// The problem

A composites maker's QC results, batch yields, customer complaints and supplier lots sit in two ERPs and the lab's spreadsheets, and the monthly quality review is a week of copying.

// What we would build

A data warehouse that takes in QC results, batch records, complaints and supplier lots from both ERPs and the lab, and shows dashboards for plant managers and the quality team: yields by line, results against specification, complaints by product and supplier. We host and run it for you in a Canadian region.

// What is in it

// Stack

• PostgreSQL
• dbt
• ERP APIs
• Next.js
• Apache Superset
• AWS Canada

// The problem

A chemical producer's lab runs on an Access database with Excel macros built fifteen years ago, its author has left, and it no longer opens on the new laptops.

// What we would build

An architecture review of the old lab system first, with a written plan, then a web-based lab system that keeps every sample, specification and result: samples logged and labelled, results entered or imported, certificates of analysis issued from the record for your quality lead to approve, and years of history moved and checked.

// What is in it

// Stack

• Next.js
• PostgreSQL
• ERP API
• PDF generation
• AWS Canada

// The problem

A resin producer sells recycled-content grades to compounders who sell to brand owners, and each party asks for proof of the recycled share, passed along as PDFs that can be edited or counted twice.

// What we would build

A shared record where the producer records each lot's recycled content with its lab result, and each sale passes that share on to the buyer, on a private permissioned ledger each company runs a node of. A signed certificate often does this job; the ledger helps when several companies share one record and a share must never be claimed twice. Claims made to the public follow the rules your advisers set.

// What is in it

// Stack

• Next.js
• Hyperledger Besu (private, permissioned)
• PostgreSQL
• ERP API
• AWS Canada