Laboratory & analytical data
Structured capture of instrument output (GC, HPLC, ICP, spectroscopy), sample metadata, units and detection limits — validated at the point of entry so every result is traceable to its batch, method and analyst.
Chemical Data
The "Chem" in ChemData
The practice began with chemical and analytical data. Chemistry demands precision: units, detection limits, sample lineage and specifications all matter. The same discipline now underpins our work in energy, commodities and capital markets — and it still starts here.
What we work on
Chemical informatics
Compound and material registries, structure identifiers (SMILES, InChI, CAS), property tables and specification limits modelled as type-safe data contracts rather than loose spreadsheets.
Process & industrial data
Plant historians, yields, mass balances and quality parameters joined into one governed PostgreSQL model, ready for dashboards, process optimisation and costing.
Regulatory & safety information
Safety data sheets, REACH / CLP classifications and internal compliance rules organised for retrieval-augmented AI that cites its sources and supports — never replaces — expert review.
Case studies
Representative engagement patterns based on the practice's chemical-informatics work. Client details are generalised; outcomes describe the system delivered, not guaranteed results.
Case study01 · Analytical laboratory
From scattered instrument exports to a traceable results database
Challenge
A laboratory team receiving analytical results as CSV and spreadsheet exports from several instruments, with inconsistent units, sample IDs typed by hand, and no reliable audit trail when a result was corrected.
Approach
- Pydantic v2 data contracts for every result: units, method, detection limits and sample lineage validated before storage.
- Automatic unit normalisation and flagging of out-of-specification or below-detection values.
- PostgreSQL schema with versioned corrections, so the original and amended values are both preserved.
Outcome
One trusted source of results, rejected records explained at entry rather than discovered months later, and reports that can be traced back to the raw instrument file.
Case study02 · Materials & industrial processing
Linking process conditions to product quality
Challenge
Process conditions, raw-material certificates and final quality tests stored in separate systems, making it hard to explain why one batch met specification and another did not.
Approach
- A batch-centred data model joining feedstock, process parameters and QC results.
- Deterministic yield and mass-balance calculations kept separate from any AI layer.
- Interactive analytical dashboards for comparing batches and spotting drift.
Outcome
Engineers can follow a batch end to end, compare conditions side by side, and base process decisions on reconciled numbers instead of manual lookups.
Case study03 · Chemical compliance
Governed AI assistant for safety and regulatory documents
Challenge
Staff spending hours searching safety data sheets and regulatory guidance, with a risk of acting on outdated or misremembered information.
Approach
- Retrieval-augmented generation over a curated, version-controlled document library.
- Every answer cites the source document and section; questions outside the library are declined.
- Compliance triage that routes uncertain cases to a qualified reviewer.
Outcome
Faster, sourced answers for routine questions, with a clear boundary: the system supports expert judgement and is not a legal or regulatory determination.
Case study04 · Chemicals meets commodities
Feedstock prices and chemical margins in one view
Challenge
Commercial teams needing to connect energy and feedstock price movements to the economics of chemical products, using data that lived in separate market and plant systems.
Approach
- Validated market-data ingestion for energy and feedstock prices.
- Deterministic margin and netback calculations with explicit assumptions.
- Scenario views showing how price shifts flow through to product economics.
Outcome
A shared, auditable picture linking chemical data with energy economics and commodity markets — the same thread as the practice's strapline.
Have chemical data that needs structure?
Tell us about your laboratory, process or compliance data and what a trustworthy system would look like for you.
Start the Conversation