Industrial research · Made Smarter Innovation
materiAI

The Material Intelligence Platform

Know your polymer before it enters production

The first affordable, edge-deployed, multi-modal spectral classification system engineered for UK SME polymer manufacturers and reprocessors. Real-time identification of six common polymers with a >95% accuracy target — in a single unit under £25k.

<500ms
Edge inference
>95%
Accuracy target
6
Polymers classified
~6 mo
Typical payback

Validated in live manufacturing environments with

Helix8 LtdPlaswire LtdBorder Plastics Recycling

The problem

A quality-control gap that costs SMEs every batch

Around 180 UK SME reprocessors have no viable option between manual sorting and industrial systems built for material recovery facilities — leaving them exposed to waste, rejected batches, and looming compliance deadlines.

3–8%

Batch rejection from contamination

Contaminated feedstock drives production waste, energy-intensive reprocessing, and material that ends up landfilled.

70–85%

Manual sorting accuracy

The only realistic option for most SMEs — reactive, labour-intensive, and unable to reliably catch degraded or black plastics.

£100k–£500k

Cost of industrial NIR systems

Designed for large MRFs, these are far out of reach for the ~180 UK SME reprocessors who need verification, not high-speed sorting.

The solution

One unit that verifies material at goods-in

materiAI fuses two sensing modalities with edge machine learning to verify polymer type and contamination before material enters production — then feeds an immutable audit trail.

Multi-modal sensor fusion

NIR spectroscopy (900–1700 nm) for molecular fingerprinting, combined with multispectral imaging (400–1000 nm) for surface and contamination features — catching cases that defeat single-sensor tools.

Edge-deployed ML

Lightweight CNN models, optimised with quantisation and pruning, deliver classification in under 500ms on a Jetson Nano-class device — fully offline, at production-line speed.

Rapid grade adaptation

Transfer-learning architectures adapt to new polymer grades with up to 80% less retraining data, so the system keeps pace with changing feedstock.

Handles the hard cases

Reliable identification of black plastics, degraded polymers, and multi-layer contaminants — the materials that undermine conventional QC.

Syncs with Helix8's platform

Works offline yet synchronises with Helix8's AWS platform for GS1 EPCIS 2.0 traceability, Digital Product Passports, and CO₂e calculation per kg processed.

Built for SME production lines

A single unit under £25k that integrates into existing lines without major infrastructure change — verification and documentation, not just sorting.

Six target polymers

Classified in real time — including the difficult ones

Molecular fingerprint plus visual signature means black, degraded, and multi-layer materials are handled, not skipped.

HDPE

High-density polyethylene

PP

Polypropylene

PET

Polyethylene terephthalate

PS

Polystyrene

PVC

Polyvinyl chloride

ABS

Acrylonitrile butadiene styrene

How it works

From feedstock to auditable record in one pass

01

Scan at goods-in

Material passes the sensor head. NIR and multispectral data are captured together in a single pass.

02

Classify on the edge

Fused CNN models identify polymer type and flag contamination in under 500ms — no cloud round-trip required.

03

Act on the result

Operators segregate incompatible materials, adjust processing, or reject genuinely contaminated feedstock before it causes failures.

04

Document automatically

Verified results feed Helix8's platform — immutable Digital Product Passport records, traceability, and CO₂e analytics.

Value & ROI

Payback in around six months

A manufacturer processing 1,000 tonnes a year and losing 5% to contamination is losing roughly £75k. Early detection can recover £50k+ annually — with contamination-driven rejection falling from around 5% to below 1%.

  • Recover £50k+ per year through early contamination detection
  • 3–5% throughput gain from reduced downtime and rework
  • Premium pricing on certified recycled-content material
  • Hardware from £8–12k plus SaaS at £700–1,000/month

Illustrative annual impact

Contamination loss avoided£50k+
Batch rejection5% → <1%
Material saved (2 trial sites)~50,000 kg/yr
System payback~6 months

Figures are illustrative, based on Helix8's project modelling at current rPET prices. Actual results vary by site, feedstock, and throughput.

Compliance & traceability

Ready for PPWR and Digital Product Passports

materiAI closes the loop: automated material verification at goods-in feeds immutable, standards-based records — giving SMEs a first-mover advantage ahead of mandatory EU PPWR and DPP rules from 2027.

GS1 EPCIS 2.0

End-to-end material traceability across the process.

JSON-LD DPP

Digital Product Passport records generated automatically.

Energy & CO₂e

Real-time monitoring and CO₂e calculation per kg processed.

Audit-ready

Verifiable evidence of recycled content for brand-owner customers.

The consortium

Built and validated with UK manufacturers

A 12-month industrial research programme integrating multi-modal classification into a live traceability platform and validating it in real production environments.

Helix8 Ltd

Technology developer & project lead

Operates the AWS IoT platform with full material traceability and Digital Product Passport capability already deployed at Plaswire.

Plaswire Ltd

Validation site 1

Mixed HDPE/PP reprocessing environment — gains real-time feedstock verification at goods-in.

Border Plastics Recycling

Validation site 2

Pellet and extrusion manufacturing — verifies incoming recycled pellet purchases against off-spec feedstock.

Get started

See materiAI on your feedstock

Tell us about your material and current QC process. We'll arrange a demo or a site assessment and show you the numbers for your line.

contact@helix8.io

United Kingdom · A Helix8 Ltd innovation

We'll only use your details to respond to this enquiry.