Back to Cases

QA/QC

Assay Guard: laboratory QA/QC review

Reconcile what you sent to the laboratory with what came back, judge every batch against your own versioned QA/QC rules, and record the decision so it survives an audit. The system recommends; a person disposes.

Request a Free Session

Demo with made-up data · In development

Who it is for

Built around one practical workflow first

  • Exploration, resource and grade-control geologists, database geologists and QA/QC coordinators.
  • Technical managers who sign off Table 1 and Item 11 text and need every number traceable to a batch and a rule version.

Workflow pain

Where the work usually slows down

  • The dispatch register knows which samples were controls and duplicates. The certificate does not, and the two never meet in one place.
  • Spreadsheets lose track of which failure was re-assayed and whether a failed standard was in fact a swapped one.
  • Every reporting period, the QA/QC narrative is rebuilt by hand from charts nobody can reproduce.

Implementation design

What Eleticle builds and controls

What is built

  • A dispatch register and certificate import that reconcile expected against received samples, including stacked-header, multi-section and re-issued certificates.
  • CRM, blank, duplicate, sequence, frequency and umpire rules in versioned rule sets that become immutable when activated, with control charts and a multi-CRM plot.
  • Failure envelopes with a system recommendation, an audited human disposition, re-assay requests, and per-batch, period and Table 1 / Item 11 reports.

What data is used

  • Your dispatch register, CSV and XLSX laboratory certificates, umpire certificates and the laboratory’s own QC report, uploaded by your team.
  • Your reference materials with certified values and dispersions, your reporting limits and your acceptance limits. No universal cut-off is assumed.

What AI may do

  • Explain what a rule measured and summarise failed and not-evaluable items.
  • Describe a dispatch reconciliation and draft a query to the laboratory.
  • Extract candidate values from a CRM certificate for a person to confirm.

What it does not do

  • Never states that data is fit for resource estimation.
  • No direct LIMS or laboratory API connection; certificates arrive by upload.
  • No universal QA/QC limits; every limit comes from your activated rule-set version.
  • Stored laboratory accreditation text is not independently verified.

Human review

The system recommends. A person disposes.

The assistant is scoped around clear boundaries: what it can read, what it can prepare, what must be reviewed, and what should stay manual.

An organisation administrator records every disposition, with a follow-up owner and date where a re-assay is pending.
The recommendation and the disposition are separate fields on every screen; nothing copies one into the other.

Before and after

Example workflow change

Before

A certificate arrives by email, a geologist eyeballs the standards in a spreadsheet, and the failed batch, its re-assay and the final decision live in three different files.

After

The certificate is reconciled against the dispatch, every control is judged by a named rule version, the failure envelope is disposed on the record, and the Table 1 text comes from the same data.

More workflows

Industrial operations

Private AI for industrial operations and technical workflows

Testwork

Testwork Atlas: testwork records with provenance

Metallurgy

Met Daily: the daily metallurgical review, signed

Comminution

Grinding & Classification Monitor: what the circuit did

Start with one free session

Share the workflow you want to improve. Eleticle will help identify the safest useful first step.

Request a Free Session