Honey authenticity
Adulteration detected with handheld NIR and PCA–LDA classification.
NIR quality control for every material
Efficient QC for anything with an NIR signature.
Food, pharma, agri, plastics or chemicals: if a material has a near-infrared fingerprint, metaspeQ can check it. Scan it as it is, with no chemicals and no sample prep, and get results in under a minute, with a reason to trust every one.
No reagentsNo sample prepFull audit trail
FoodPharmaAgriPlasticsChemicalsHerbal
METASPEQ Studio · Blend line 1 · paracetamol-blend v1.0.0
Simulated
WITHIN RANGE
Predicted 100.4 % label claim ± 2.1
Hotelling T² 0.01× limit · Q residual 0.29× limit · identity 1.0000
Audit A-2609-0142 · sha256 3f9c…e1a7
Time to result < 1 min
Backed and recognised
Incubated at IIT Delhi IHFC
Grant-backed by IIMCIP
Member, IDMA
Patent No. 557176
Top 5 finalist, TiE Kolkata ThriveX 2026
Meet the analyser
A compact NIR spectrometer that fits on any QC bench.
Place the sample, press acquire, and METASPEQ Studio does the rest: dark and reference checks, prediction with the right model, a domain check and an audit record.
- Detector
- Hamamatsu NIR detector, 256 pixels
- Samples
- Solids, powders and liquids, as they are
- Scans
- 1 to 64 replicates per sample, median reported
- Connects
- Outbound-only edge agent on the instrument PC
The problem
Most batches aren't late because of production. They're waiting for QC.
Every raw-material batch goes through the same slow chain. None of its steps makes the product better.
Wet-chemistry QCtoday, at most plants
Sample prep by hand
Reagents and solvents
Trained analyst
Lab queue
Batch waits
metaspeQat the point of use
Scan the sample as it is
Result in under a minute
How it works
From spectrum to a result QA can trust, with every step on the record.
Instrument
Guided dark, reference and sample scans, with 1 to 64 replicates per sample.
Edge agent
A small service beside the analyser. It only calls out, so no port is opened on the instrument PC.
Prediction
The server validates the spectrum, predicts with the model's own preprocessing and checks the calibration range.
Audit trail
The original spectrum and every action are stored in a hash-chained record an auditor can verify.
Dashboard
Results arrive live in the browser, and drift monitors watch every one.
QC from anywhere
The scan happens at the plant. The decision can happen anywhere.
The operator scans at the bench. METASPEQ Studio checks the result, records it and sends it to QA. QA reviews and signs from any browser, at home, on the road or in the air.
Plant floor · Blend line 1
1. Scan at the bench. Results in under a minute.
METASPEQ Studio · Microsoft Azure
- Input checks passed
- Inside the calibration range
- Prediction recorded
- Audit entry chained
- Sent to QA for signature
2. Checked and recorded. Every step on the audit trail.
QA, wherever they are
B-2609-02 · example
Paracetamol blend
WITHIN RANGE
Sign
3. Signed from home. Close out the evening batch from the kitchen table.
Signing always asks for a fresh Microsoft sign-in, wherever you are, and follows your own release SOPs. The analyser, the sample and the operator stay at the plant.
METASPEQ Studio
No number without a reason to trust it.
A spectrometer gives a number in seconds. The hard part on a production floor is trusting that number and proving who relied on it.
Refused
Bad input
Corrupt, incomplete or out-of-range spectra are rejected before prediction, and kept as evidence.
No value
Unknown sample
A sample outside the model's calibration range is marked OUT OF DOMAIN. It gets no number, not a guess.
Built in
Signed models
Each model carries its own preprocessing and is checksummed and signed, so there's no setting to get wrong.
Every change that matters needs a second signature
- ANALYSTRequests the change. The exact request is fingerprinted.
- QAReviews exactly what they're about to sign.
- QASigns after a fresh sign-in.
- ANALYSTExecutes. The fingerprint is checked again first.
Requester ≠ approverSingle useTime-limitedAdmins can't sign
Covers model promotion, batch release and process interventions. Studio supports your validation (IQ/OQ/PQ); validation, SOPs and training stay with your quality team.
Agents propose. Humans approve.
< 1 min
From scan to result
up to 90%
Lower running costs. No reagents or consumables per test
50–80%
Lower purchase price than legacy NIR systems
20+
Peer-reviewed NIR studies behind the models
Analysers
One platform, from the QC lab to the production line.
Every analyser runs the same calibration models and connects to METASPEQ Studio.
Watch the demoYouTubeNEXGA DS01
Desktop, solids
Powders, granules and tablets, in the lab or beside the line.
Available now
NEXGA DL01Desktop, liquids
Solvents, syrups and liquid raw materials.
Available nowHandheld
Field unit
Screening at the warehouse door and in the field.
Planned 2028In-line PAT
Production line
Continuous monitoring inside the process.
Planned 2029
Software
METASPEQ Studio
Governed QC: measurement, prediction, automated model building, signed approvals and a verifiable audit trail.
Explore Studio → ServiceNIR validation support
Help with method development, calibration and your IQ/OQ/PQ, plus training after deployment.
Read the validation guide (PDF) →Applications
One platform, every material with an NIR signature.
The same analyser and the same METASPEQ Studio work across industries. Only the calibration model changes, and each proven application comes with a method note you can download.
Dyes and fillers in spices
Sudan dye, metanil yellow and starch adulteration in turmeric, screened from a scan.
Tablets and counterfeits
Paracetamol content in whole tablets, and spectral fingerprints that flag suspect products at receiving.
Diethylene glycol in cough syrup
A fast, portable screen for DEG contamination in glycerin-based syrups.
Tea grading
Polyphenol, caffeine and grade estimates that support CTC tea auctions.
Lignin in jute
Fibre characterisation by PLS regression, presented at ICNIRS 2025.
Herbal raw materials
Quinine in cinchona bark, andrographolides, and ginger traceability by fingerprint.
Polymer identification
Sorting and incoming checks for polymers and packaging films. Ask us about a feasibility study.
Raw-material identity and moisture
Identity checks and moisture for solvents, powders and intermediates. Ask us about a feasibility study.
Your material isn't listed? If it has an NIR signature, send us a few samples and we'll tell you whether a model is feasible.
The science
Built on two decades of NIR research.
Anchored by Prof. Rajib Bandyopadhyay of Jadavpur University, our team has published 26 peer-reviewed NIR studies, from adulteration in spices to herbal raw materials. Every one links to its public DOI page.
- TurmericDetecting Sudan dye I adulteration with NIR and support vector machines
- GingerRapid quality and traceability of ginger powder from Northeast India by spectral fingerprinting
- HerbalsQuality of cinchona, andrographis and black pepper with a portable NIR device
- TeaPredicting polyphenol content in fresh tea leaves with machine learning
- PulsesDetecting metanil yellow adulteration in pigeon pea
ICNIRS 2025 · Sapienza University of Rome · 15–19 Sep 2025
Five contributions at the world's main NIR conference
- OralParacetamol in intact tablets with NIR and artificial neural networks
- OralLignin in jute by PLS regression with SNV and second-derivative preprocessing
- PosterPortable NIR screening for diethylene glycol in cough syrup
- PosterQuinine in cinchona bark with a transformer model on NIR spectra
- PosterNIR for grading and pricing CTC black tea
NIR-Sight webinar series
Hosted by metaspeQ ahead of the 9th Asian NIR Symposium (ANS-2024)
Monthly sessions with NIR and chemometrics researchers from India and abroad.
Analysis of spectroscopic data with ASCAProf. Federico Marini, Sapienza University of Rome
Hyperspectral NIR imaging in agricultureDr. Rabi Narayan Sahoo, ICAR-IARI, New Delhi
R&D on NIR spectroscopy in JapanAgriculture and forestry
NIR-Sight webinar 5Prof. Dmitry Kirsanov
Our journey
From a university lab to the plant floor.
metaspeQ started as NIR research at Jadavpur University. Nine years on, it's a desktop analyser and software at work in customer labs, backed by incubators, industry programmes and peer-reviewed science.
- 2017Research begins
NIR and chemometrics R&D at Jadavpur University
- 2020Company founded
Ayudyog Private Limited incorporated, DPIIT-recognised
- 2022Prototype I
First in-house NIR spectrometer
- 2023Patent granted
Patent No. 557176
- 2024Prototype II
Field trials, and a booth at the 9th Asian NIR Symposium
- 2025MVP and first orders
Desktop analyser and Studio in customer labs
- NextHandheld and in-line
Field unit, then continuous monitoring on the line
Incubated, with ₹50 lakh in funding
The Technology Innovation Hub at IIT Delhi incubated Ayudyog and backed the build-out of the analyser.
GSK Pad Up, 15 of 200
One of 15 startups chosen from 200 applicants for GSK's startup programme.
Young Innovator award
Recognised at the Tata Social Enterprise Challenge for chemical-free, prep-free quality testing.
9th Asian NIR Symposium
Live demos for NIR researchers from across Asia, after a year of NIR-Sight webinars.
Cinchona bark, in the field
Quality checks for cinchona bark, with a certificate of appreciation from the Directorate of Cinchona and Other Medicinal Plants.
Paracetamol in intact tablets
An industrial collaboration with Mendine Pharmaceuticals, measuring tablets as they are.
- Top 5 finalist, TiE Kolkata ThriveX 2026
- Speaker, CII Eastern Region Annual Meeting 2025
- Live demo at the IDMA workshop on revised Schedule M
- Orders and partners: Spectramind Tea, Tea Mech, Ultra Cure, Bengal Herbs
- Government projects with C-DAC Kolkata under MeitY
- Five contributions at ICNIRS 2025, Rome
Paid pilots
See it work on your own materials.
A pilot runs on one material you test often, with your lab method as the referee.
Pick one material
A raw material, a blend or a finished product.
We calibrate
Using your own samples and your lab's reference values.
Run side by side
Your current method stays the referee.
Review the report together
We quote hold-out error, measured on samples the model never saw.
You decide
Scale up, or stop with a clear answer.
Team
Built by people who've run QC labs.
CEO
Soumitra DasMahapatra
8 years in pharma quality
COO
Uday Singh
8 years in pharma manufacturing
CTO
Dr. Dilip Sing
9 years in instrumentation
CPO
Dr. Subhadip Banerjee
8 years in chemometrics
Director
Prof. Rajib Bandyopadhyay
35 years as a professor