Why it matters
Without AI, mis-labelling or cross-contamination can trigger processing delays, context mismatches, and safety concerns during handling of samples, creating backlogs and risk in post-excavation reporting.
From design to safety
Design
AI supports sampling design and planning
AI ingests desk-based assessments, WSI, sampling strategies, petite bulk sample counts, known horizons, and past flora data to optimize flotation and processing workflows before trench work begins, aligning with HER and CIfA standards; it defines target recovery rates, matrix tolerances, and QA checkpoints.
- bulk samples
- flotation tanks
- sample registers
Build
Automated setup and in-field routing
AI reads field records, sample logs, water quality sensors (pH, temp, ORP), turbidity meters, resin load targets, and reagent schedules to validate tank placement, water chemistry stability, flow rates, and staged releases while directing crews to correct stations and recording timestamps for audit trails.
- bulk samples
- flotation tanks
- residues
Operation
Real-time processing and documentation
AI tracks residues and charred plant remains from flotation outputs, cross-references with context sheets and Harris matrix, flags potential miscounts or cross-contamination, logs barcodes, and generates interim reports for daily client meetings and post-excavation records.
- flots
- charred plant remains
- residues
Maintenance
Calibration and archivist-ready records
AI monitors flotation water chemistry, screen wear, sieve apertures, agitator speeds, reagent stocks, and sample register integrity, triggering maintenance tasks, calibration logs, and ensuring archive deposition readiness with full audit trails and chain-of-custody records.
- sample registers
- flotation tanks
- flots
Safety
Buried services and contamination alerts
AI cross-references trench drawings, flotation area, and utility maps to identify hazards, while staff receive protective guidelines, PPE checks, and incident logging prompts; it also screens for cross-contamination risks and flags abnormal readings for immediate action.
- bulk samples
- flotation tanks
- sample registers
Part by part
- 1
bulk samples
- Watch for
- Mislabeling or mixing of samples leading to incorrect context assignment.
- AI action
- AI flags mismatched codes and prompts re-checking of IDs.
- 2
flotation tanks
- Watch for
- Clogging, uneven slurry, or stalled separation affecting recovery.
- AI action
- AI monitors turbidity and flow, triggers clean-out and re-balancing.
- 3
flots
- Watch for
- Underdetection of light fractions or contaminants during screening.
- AI action
- AI alerts operator to re-sieve and re-float selectively.
- 4
residues
- Watch for
- Cross-contamination between residues and controls.
- AI action
- AI flags cross-contact and prompts separate retests.
- 5
charred plant remains
- Watch for
- Degradation or loss of fragile macro-remains during processing.
- AI action
- AI orders gentle handling and alternate drying protocols.
- 6
sample registers
- Watch for
- Inconsistent entries across journals, leading to audit risk.
- AI action
- AI validates entries against barcodes and generates reconciliation report.
Worked example
Environmental sample flotation management on 2.3 ha site with medieval layers
Illustrative scenario with sample data.
During a development-led excavation, AI manages bulk samples and flotation outputs to differentiate charred plant remains from residues across multiple trenches and contexts, supporting a post-excavation assessment.
Signal in
bulkSamples=120, flotationTankTemp=22C, floatVolume=35L, residueWeight=2.8kg, plantRemains=42 fragments, sampleID=BR-2026-07-19, pH=7.2, turbidity=5 NTU, date=2026-10-14
System out
generatedProcessingQueue: BR-2026-07-19, priority=high, action=perform flotation run, update context sheets, archive deposit instructions, flag for QA review, log audit trail ID: AT-BR-2026-07-19
Human sign-off: Post-excavation supervisor approves the processing queue, confirms interpretation against WSI, and signs off on the sampling plan before HER submission.
How to roll it out
Phase 1
Weeks 1 to 2
Baseline setup of flotation units, sample registers, and initial WSI alignment with field-team onboarding and calibration of sensors.
Phase 2
Weeks 3 to 6
Integration with context sheets, Harris matrix, and single context recording templates; QA checks, quality control samples, and POS tagging established.
Phase 3
Weeks 7 and beyond
Full data flow, automated reporting, and archive deposition readiness; ongoing calibration, safety audits, and continuous improvement cycles with regulators.
Where people stay in charge
- Decisions on sample prioritization and reporting must be signed off by the project manager or heritage consultant.
- All handling and disposal steps require approval by the finds supervisor and safety officer.
- QA and archive deposition must be signed off by the post-excavation manager before HER submission.
Questions owners and crews ask
What is AI's role in environmental sample processing on site?
AI assists in tracking sample IDs, optimizing flotation sequencing, and flagging anomalies, while humans confirm processing decisions and ensure compliance with WSI and HER.
How does AI handle quality control in flotation?
AI monitors turbidity, flow rates, pH, temperature, and tank cleanliness; it raises corrective actions and logs decisions to the sample register for traceability.
Who approves the processed results before deposition?
The post-excavation manager or curator signs off on results, with the project manager ensuring all records are complete and aligned with the written scheme of investigation.
What happens if a hazardous residue is detected?
AI flags the hit; the safety officer implements containment and disposal procedures, then the finds team documents it in the context sheets and updates the HER as required.