Why it matters
Without AI, misclassification of fabrics, misdated sherds, or missed decorative signals can lead to backlogs, misinterpretation of context, and delays in post-excavation assessment and archive deposition.
From design to safety
Design
AI informs study design for pottery
AI reads pottery fabrics, sherd forms, and decoration patterns to propose fabric series groupings and draft spot-date hypotheses within the Written Scheme of Investigation. It integrates stratigraphic cues, metadata, and provenance tags for early planning decisions.
- pottery fabrics
- sherd forms
- decoration
- spot dates
Build
AI-guided finds processing workflow
AI reads incoming finds data, wires fabric series to context sheets, flags anomalies, and assigns provisional codes for single context recording. It interfaces with plan charts, week-by-week excavation logs, and sampling plans.
- fabric series
- context dating
- spot dates
Operation
Real-time classification and dating outputs
AI updates context sheets, generates plan overlays and Harris matrix links, and flags findings needing curatorial review or further sampling. It exports interim archive-ready records and dashboard summaries.
- context sheets
- plans
- photogrammetry
Maintenance
Data integrity and archive preparation
AI validates entries, reconciles fabric/decoration codes with reference collection, and flags records requiring post-excavation assessment. It verifies file naming, backup integrity, and metadata consistency across HER submissions.
- ceramic fabric reference collection
- shard form catalog
- decoration catalog
Safety
Quality control and safeguarding evidence
AI cross-checks dating signals against stratigraphic context, ensures safe handling of finds data, and supports secure archiving and HER submissions. It monitors provenance trails and access logs.
- context dating
- spot dates
- archive deposition
Part by part
- 1
Portable XRF spectrometer
- Watch for
- instrument drift or miscalibration leading to spurious elemental signals on fabric series.
- AI action
- AI monitors readings and flags drift; triggers calibration reminder and retest workflow.
- 2
Handheld optical microscope
- Watch for
- poor image resolution or misidentification of temper and inclusions.
- AI action
- AI evidences quality check and prompts retake or higher magnification imaging.
- 3
Digital calipers
- Watch for
- measurement variance beyond tolerance for form classification.
- AI action
- AI compares measurements to form templates and flags outliers for retake.
- 4
Ceramic fabric reference collection cabinet
- Watch for
- missing or unindexed fabrics hindering fabric-series assignment.
- AI action
- AI verifies codes against reference entries and requests catalog update.
- 5
Context sheet binder with bin labels
- Watch for
- mislabelled contexts or misaligned Harris matrix entries.
- AI action
- AI cross-checks context numbers with plan overlays and flags inconsistencies.
- 6
3D photogrammetry rig
- Watch for
- missing or inaccurate dimensional data for context planning.
- AI action
- AI validates spatial data, updates site model, and notifies recorder to recapture if needed.
Worked example
Catching a spot-date signal before the trench backfill
Illustrative scenario with sample data.
A 0.8 ha development site yields decorated wares; AI flags a contested spot-date within a buried gravel horizon.
Signal in
site_id: NorthPit-03, stratigraphy: [Layer A, Layer B], fabric_codes: [F1, F2], decorations: [pale slip, comb pattern], measured dims: [length: 22mm, width: 18mm], context_sheet_id: CS-127, photographs: IMG_0123-IMG_0128
System out
AI-generated provisional dating window: 210–230 CE; context link to Layer B; recommended retake of decorated sherd with higher-res imaging; report to curator.
Human sign-off: Project manager reviews and signs off, with HER liaison for potential backfill condition constraints.
How to roll it out
Phase 1
Weeks 1 to 4
Baseline data infrastructure is established, initial fabric references are catalogued, and QA workflows are defined.
Phase 2
Weeks 5 to 12
AI integration with field recording and sample processing accelerates classification and dating decisions.
Phase 3
Weeks 13 to 20
Archive workflows and reporting templates are validated, and post-excavation assessment planning begins.
Where people stay in charge
- All dating decisions beyond provisional spot dates must be reviewed and signed off by the licensed archaeologist or HER authority.
- AI-derived identifications requiring legal provenance must be auditable through the chain of custody and curator sign-off.
- Decisions affecting sensitive heritage constraints require documented approval from the planning archaeologist and client heritage advisor.
Questions owners and crews ask
What is AI pottery sherd classification used for in excavation planning?
AI pottery sherd classification supports fabric series assignment, decoration analysis, and provisional spot dating, feeding into the Written Scheme of Investigation and future archive deposition.
Who signs off spot-date decisions in the field?
The qualified archaeologist, under the supervision of the project manager, signs off provisional dates and coordinates with the local authority planning archaeologist.
How does AI handle fabric series and context dating?
AI maps sherd fabrics to reference collections, links dates to stratigraphic context, and flags anomalies for curator review and post-excavation assessment.
What records are generated by this AI system?
AI produces context updates, dating notes, fabric-series assignments, plan overlays, photogrammetry outputs, and archiving metadata suitable for HER submissions.