Our technology consists of a three-part system: 1. a sensor or spectrometer, which may be an Enterprise Sensor, a Preemie Sensor, or another type; 2. our cloud-based patented Artificial Intelligence (AI) engine, called SpecAI, and 3. our mobile app, SpecData, which is used to scan samples, and feed SpecAI to train or execute Machine Learning (ML) models.
Our technology helps monitor, in real time, events of fraud, illicit trade, and intentional or unintentional contamination. We also help manage sample variability by offering rapid and in-the-field analysis of samples, thereby supporting quality control systems. A quick scan, and within seconds your smartphone or tablet displays the results you need to make an informed decision.
SpecAI is an end-to-end AI platform for easy, fast development and deployment of ML models from spectral data. Our patented AI platform was developed with a user-friendly drag and drop interface, where one can select widgets and the connections between them to quickly perform data treatments, visualize data, build ML models, perform cross-validation, and send it to the next widget for further processing. In a few clicks, the user can start training a model by using an extensive library of ML algorithms.
SpecAI is ideal for collaborative work: it automatically synchronises with the cloud server to ensure data is stored in a secure and consistent way across all users, and that data collected anywhere is accessible to administrators and data scientists. Administrators can manage users, set up protocols for data collection, check the data collection process, and publish calibration models for use by sensors in remote locations on the Internet.
SpecAI is sensor agnostic: researchers can upload data from any spectrometer, or by using the SpecData mobile app with Tellspec’s Enterprise Sensor or Preemie Sensor.
SECURITY
Store collected data in a secure company-wide cloud database, register users and manage their access.
COLLABORATION
Multiple users can work on the same dataset simultaneously from anywhere, and share data and models in real time.
EXTENSIVE DATA VISUALIZATION
Visualize and explore data, such as outliers, variable distributions, or data clusters without having to download it to a local computer.
COMPREHENSIVE DATA PRETREATMENTS
Access in an intuitive way a comprehensive set of data pretreatments commonly used in chemometrics.
EASY AND FAST ML MODEL BUILDING
Access a growing library of ML algorithms and choose from regression or categorical model types, as well as perform hyperparameter optimizations with several algorithms at the same time.
POWERFUL EXPERIMENTATION
Flexible architecture that helps you experiment, build, train, and deploy state-of-the-art ML models.
SENSOR AGNOSTIC
Upload data from any spectrometer or by using SpecData mobile app with Tellspec’s Enterprise Sensor or Preemie Sensor.
SCALABILITY
Effortlessly increase your computational power by adding servers to your cloud architecture.
DIVERSIFICATION
Export the results in a variety of formats, like csv, json, xlsx, or pdf, and get results from a variety of processes.
TECHNICAL SPECIFICATIONS
Server hosted at AWS. Web browsers supported: Google Chrome, Mozilla Firefox, Microsoft Edge, Opera, and Safari.
Windows and MacOS compatible
The SpecData system uses either the Enterprise Sensor or the Preemie Sensor to collect spectral data that later is used in the development of calibration models.
The SpecData software has two components: the desktop application, which is used by an administrator to set up users’ access and track the data collection process, and the mobile application, that is used by a team and allows the collection of spectral data along with label data from laboratory analysis. The data collected using SpecData is stored in our cloud and can be easily accessed by SpecAI for training and deploying ML models.
Tellspec’s two sensors are near-infrared spectrometers based on the NIRscan™ Nano design from Texas Instruments and their innovative DLP® technology.
Technical Specifications
Enterprise Sensor
| Spectroscopy | Near-infrared, reflective |
| Size | 82 mm (L) x 66 mm (W) x 45 mm (H) |
| Weight | 136 grams |
| Cuvette Holder | N/A |
| Illumination | Two integrated tungsten halogen lamps, 0.7 W each |
| Spectral Range | 900 nm to 1700 nm |
| Monochromator | Plane grating and TI DMD DLP2010NIR, 854×480 pixels |
| Detector | Single element 1 mm InGaAs detector, uncooled |
| Resolution (FWHM) | 10 nm |
| Signal to noise ratio | 5000:1 (typical) |
| Scan Pattern | 256 Hadamard pattern columns |
| Total Scan Time | 6 sec |
| Power | USB (500 mA @ 5 V) to charge integrated Li-ion battery (3.7V/1000mAh) |
| Battery Life | 300 cycles |
| Battery Charging Time | 2.5 hours (fast) and 8 hours (standard) |
| Battery Usage | ~1000 scans from full charge to empty |
| OS Support | Android, iOS |
| Connectivity | Bluetooth Low Energy (wireless) |
| Supported OS’s | iOS, Android |
| Data Format | Internal, with CSV output |
| Ambient Temperature, operating | +10°C to +40°C |
| Ambient Humidity, operating | 10 % to 80 % relative humidity, non-condensing |
| Colors | Black |
Preemie Sensor
| Spectroscopy | Near-infrared, transmissive |
| Size | 125 mm (L) x 73 mm (W) x 49 mm (H) |
| Weight | 233 grams |
| Cuvette Holder | 12.5 x 12.5 x 35 mm, Z-dimension 15 mm |
| Illumination | One integrated halogen tungsten lamp, 0.7 W |
| Spectral Range | 1350 nm to 2150 nm |
| Monochromator | Plane grating and TI DMD DLP2010NIR, 854×480 pixels |
| Detector | Single element 1 mm InGaAs detector, uncooled |
| Resolution (FWHM) | 10 nm |
| Signal to noise ratio | 5000:1 (typical) |
| Scan Pattern | 256 equally-spaced columns |
| Total Scan Time | 16 sec |
| Power | USB (500 mA @ 5 V) to charge integrated Li-ion battery (3.7V/1000mAh) |
| Battery Life | 300 cycles |
| Battery Charging Time | 2.5 hours (fast) and 8 hours (standard) |
| Battery Usage | ~1200 scans from full charge to empty |
| OS Support | Android, iOS |
| Connectivity | Bluetooth Low Energy (wireless) |
| Supported OS’s | iOS, Android |
| Data Format | Internal, with CSV output |
| Ambient Temperature, operating | +10°C to +40°C |
| Ambient Humidity, operating | 10 % to 80 % relative humidity, non-condensing |
| Colors | Pink/white |
