DNXT.AI turns diamond polishing — today artisanal, variable and labour-intensive — into an intelligent, standardised and traceable industrial process. A closed-loop robotic cell fusing AI, precision robotics, computer vision, haptic sensing and blockchain.
Eligible investment
Project duration
Target maturity (7–8)
Polishing precision < 1µm
Polishing converts a rough diamond's unique properties — hardness, crystal structure, purity — into brilliance, symmetry and commercial value. It is also the most complex and critical phase: the final shape depends on physical and chemical characteristics and the intended cut, demanding constant re-evaluation.
Today it remains heavily dependent on specialised manual labour — low standardisation, high risk of fractures and material waste. Operator experience directly drives speed, precision and value, which amplifies variability.
Material lost per stone with conventional methods — no dynamic pressure or angle control.
Quality control is still manual or semi-automated, with no early micro-defect detection.
Traceability is siloed and vulnerable to integrity failures and record manipulation.
A closed-loop robotic cell that adjusts pressure, angle and speed in real time, runs inline quality control (vision + haptics) and guarantees end-to-end digital provenance per stone through tamper-evident records. It closes the gap between today's planning/measurement modules and real-time adaptive polishing — introducing an intelligent decision module that autonomously updates operating parameters from continuous visual and tactile feedback.
Ci01–Ci05 — the disruptive building blocks that position DNXT.AI clearly beyond today's diamond-polishing technology.
A fully automated robotic arm with tactile sensors (pressure, vibration, texture) and precision actuators, driven by adaptive AI that adjusts pressure, angle and speed in real time in a closed loop.
High-resolution cameras, structured lighting and deep-learning models detect, segment and classify fractures, inclusions and surface defects inline — before they compromise the cut.
A Reinforcement Learning module defines, optimises and updates 3D polishing trajectories dynamically, from each stone's shape, composition and physical properties, with real-time visual and tactile feedback.
AI dynamically controls angle, pressure and speed to cut waste and energy — where conventional methods lose 10–20% of material per stone with no dynamic control.
A public, auditable blockchain records every processing step — rough stone to finished gem — binding machine data (force/vision, recipes, parameters, timestamps) to tamper-evident records.
A continuous feedback loop where every component informs the next — and every step is recorded.
High-res microscopy, cross-polarised & IR imaging, 3D models and chemical composition — normalised and interoperable (OPC-UA / MQTT).
Diamond detection & edge delineation, stress-region segmentation and facet-roughness estimation feed the decision core.
Rule-based projection then RL optimises 3D trajectories and adaptive force profiles under pressure, angle and geometry constraints.
Hybrid force/position and impedance control execute light, controlled contact — a programmable spring-damper that protects the stone.
Every critical operation, quality check and adjustment is written to a tamper-evident public ledger for end-to-end provenance.
Intelligent robotic system with AI & haptic feedback — polishing under 1 µm, real-time parameter adjustment.
Computer vision & deep learning for fracture detection, segmentation and classification.
Reinforcement learning for dynamic 3D polishing trajectories, adapted per stone.
Blockchain-based traceability & transparency — public, auditable, end-to-end.
Interoperability & industrial integration — 100% compatibility with Leica, Sarine, OGI.
Sustainability — ≥15% less diamond loss, ≥10% less energy per unit processed.
Productivity & global competitiveness — +40% per-operator productivity, +30% capacity.
A phased, cumulative plan taking the consortium from industrial research to validation in a relevant environment (TRL 7–8).
Technical-scientific survey of AI, computer-vision and robotics algorithms for real-time, reliable, safe diamond handling.
Data architecture, interoperability with existing systems and the blockchain traceability model; acquisition and categorisation of multimodal data.
Detection & edge delineation, stress-region segmentation and surface-roughness/defect prediction across multiple facets.
Kinematic/dynamic modelling, hybrid force/position & impedance control, and the state-machine software architecture.
Rule-based trajectory projection and reinforcement learning to optimise paths and adaptive force profiles in real conditions.
End-to-end beta prototype integrating vision, AI and robotic control; simulation → real production; blockchain security validation.
Project website & identity, scientific papers, master's & PhD theses, trade fairs and a public demonstration event.
Governance, risk management, IP protection & technology watch, and the exploitation & sustainability roadmap.
Global diamond market, 2024
Projected market by 2032 (CAGR 4.5%)
Targeted turnover growth, year 1 post-project
Annual revenue from tech licensing
Internal: ~30% less average polishing time, ≥15% less material loss, up to 10% less energy per unit — raising facet consistency (≥25%) and margins.
External: licensing, supply of robotic equipment/cells and integration services for cutters and processors — starting in India (Surat) and Africa (Botswana, Angola, South Africa), then Europe and North America.
DNXT.AI contributes directly to Europe's green and digital transition and to the STEP priority technologies:
Artificial intelligence (vision & RL), advanced robotics, industrial data integration (OEE-ready) and interoperability (OPC-UA / MQTT).
Material and energy efficiency gains, with significant reduction of waste and rework.
Closed-loop robotic cell with force control and predictive endpointing, raising TRL to 7–8 for EU industrial scale-up.
Blockchain for provenance and tamper-evident process audit, reinforcing ESG transparency and value-chain confidence.
Lead · SME · Technology taker
A Porto-based microenterprise founded in 2022, pioneering the digitisation, automation and full traceability of the diamond value chain. Promoter of the DNXT RWA Marketplace (Portugal 2030), it brings deep market knowledge, real-world validation, integration and commercial exploitation.
R&D lead · ENESII
A public university in Covilhã with 18 research units, including NOVA LINCS UBI and the Instituto de Telecomunicações — both rated “Excellent” by the FCT. UBI leads R&D in computer vision, robotics and reinforcement learning.
27+ years in precious stones — extraction and cutting to global trade — with blockchain and digital expertise.
FEUP MSc. Mechanical & structural design, integration and experimental validation of the robotic prototypes.
Blockchain, Web3 and RWA tokenisation architecture for the traceability module.
Compliance Director — IP strategy, GDPR and technology contracts.
Associate Professor, 50+ papers in computer vision, AI and biometrics. Leads UBI's R&D.
100+ publications; heads the SOCIA Lab. Supervises CV, deep learning and RL development.
160+ papers in robotics, automation, energy efficiency and Industry 4.0 systems.
Deep learning, computer vision and pattern recognition for industrial and biometric contexts.
Robotics, sensor fusion and force-adaptive manipulation with 6D perception.
Diamond NXT's innovation trajectory is supported by the European Union through the COMPETE 2030 programme. The DNXT RWA Marketplace operation (COMPETE2030-FEDER-01452600) advances blockchain-based diamond trading — the foundation on which DNXT.AI's end-to-end provenance is built.
