top of page

AI and Analytics Solutions

Aligning machine intelligence to have smart, reliable and increasingly accurate data

paginaprincipal.jpg
Saiba mais

Discover Analytics and Artificial Intelligence solutions 

Analytics

Master Data Management (MDM)

Robot Process Automation

data quality

Biometrics and computer vision

Natural Language Processing

Chatbot Platform

Predictive Modeling

optimization

GeoAnalytics

Data Quality

Planning, implementing and controlling activities that apply quality management techniques to data to ensure that it is fit for consumption and meets the needs of data consumers.*DMBok V2

GettyImages-1128252197.jpg

Data Quality Process

qulidaenegocios_edited.jpg

Business

DataQuality.png
qualidadeti_edited.jpg

YOU

Data Quality Objectives

qualidade02_edited_edited.png

Enable through data quality processes so customer systems have data: 

Correct, Precise, Consistent, Complete, Integrated, Adhering to

business rules and Adherents to established domains.

Dimensions of Data Quality

qualidade03.png

During the Data Qualification process, these are subject to an evaluation on each  quality dimension (or criterion). These dimensions define what will be measured from the verified.

Case: Cadastral Qualification Project

With the objective of continuously improving the quality of the data of the structuring systems of people management, the PQC works in the continuous improvement of diagnoses, graphs and quality indicators.

Benefits:

  • Allow the public manager to have qualified information for decision making

  • Improve the compliance of the information that is in the systems so that, for example, it is reflected in fair payments to employees, retirees and pensioners

  • Protect the data subject, as a duplicated data for another person, for example, can bring burdens and responsibilities for those who manage the data and also for the data subject himself

  • Allow the manager responsible for the data in the system to have tools for monitoring the quality of the data so that he can adopt timely adjustment measures, if necessary

Case: RAIS - MTE

case03.png

Serpro initiative, with the objective of mitigating project risks associated with Data Quality.

Data Qualification acts as another layer of security and validation for RAIS deliveries.

More than 200 diagnoses have already been carried out in a universe of 240 million records and the needs for remedial measures identified and applied during qualification guarantee greater security in the delivery of processing.

Âncora 2

Biometrics and Computer Vision

THEComputer visionconsists of making computers see, to gain a high-level understanding from digital images and videos. We can have several applications:

  • Object detection and classification

  • Identification /Biometryby image (Ex: facial recognition)

  • Pattern recognition and monitoring

  • Motion detection and estimation

  • 3D analysis from 2D images

THEBiometryis the analysis of physical or behavioral characteristics of people in order to uniquely identify them.

digital.jpg

Case: Passive Liveness

Case Liveness Passivo.png
  • Passive Liveness – No movement is required for validation

  • Compliance with ISO/IEC 30107-3

  • FIDO accredited laboratory confirmation letter

  • Ready for:

      _cc781905-5cde-3194-bb3b-136

      _cc781905-5cde-3194-badplay5cf-58d6

      _cc781905-5cde-3194-bb3b-136

  • Mobile and Web

  • Front and rear camera

  • Millions of monthly validations

Case: VCIT - Cognitive Validator of Traffic Violations

  • Identifies the car in the image​

  • Uses Deep Learning to classify the ​
    car by make and model (250 car makes and models)​

  • Detects the board and performs OCR​

  • Enables automation of more than 80% of tax assessment notices​

  • Hachure the windows, preventing the recognition of passengers

Case VCIT - Validador Cognitivo de Infrações de Trânsito.png

Case: Cognitive Document Validator

Cases Biometria - Validador Cognitivo de Documentos.png

Identify documents, detecting their fields and extracting information from them - making it possible to submit this information to government databases to validate the data

Case: Signature Detection

Case Deteccção de Assinaturas.png
  • Problem: High number of infraction notices without valid knowledge (approximately 60%)

  • Late identification leads to cancellation of the infraction notice

  • A computer vision model was created that runs through all the records and identifies those that do not have valid science - signature of the assessed person or witnesses.

Case: AIBIO - Biometric Ecosystem

Aibiq.png
  • Allows ingestion of large biometric databases

  • Processing and control of biometrics

  • Deduplication process - looking for possible fraud or errors in the database

  • Solution that allows you to receive a facial base and compare individuals with each other, looking for possible frauds

  • Centralized and trusted biometric database creation

  • Biographical and biometric consultation routines

Case: Biovoice

biovoz01.png

biovoice

Welcome to)

User onboarding

bovoz02.png

Speak:

 

authentication phrase

identified user

95% sure

biometric authentication

Results feedback

It is a service (API) that allows you to use thevoice recognitionas a factor ofbiometric authentication.

Predictive Modeling

Predictive modeling is a process through which a future outcome or behavior is predicted based on past data, this process is created from mathematical models and applied through Machine Learning.

GettyImages-1200758333.jpg

Case: Crime Expectation Map 

crimes.png

Application of Artificial Intelligence with the use of a predictive model capable of classifying the expectation of crime in each region of the city and according to the date and period of the day, aiming at the mainly ostensible optimization of police work.

Case: Reference Price

compras01.png
compras02.png

Solution that uses Machine Learning to forecast reference prices for materials and services and identify the best time of year to purchase these items by public agencies.

optimization

Process optimization is a solution that maps the activities performed by organizations, enabling the identification of problems such as failures, low performance and possible bottlenecks, proposing solutions to correct and measure their results.

otimizacao01.jpg

Case: CRIO - Cognitive Process Grouper

crio01_edited.png

Scope:Through unsupervised learning techniques, group similar processes. These groupings can be weighted and must be carried in the auditor's backpack according to their working capacity.

Problem:Many equal processes being judged by different auditors - rework. Priority processes enter a queue and may miss the deadline due to the amount of work.

  • Optimize process analysis

  • Grouping of similar processes
    based on heuristics

          _cc781905-5cde-3194 -bb3b-136bad5cf58d_ - Similar theme

          _cc781905-5cde-3194 -bb3b-136bad5cf58d_ - Processes from the same company, etc.

  • Clusters are given work weights

  • Auditors support workloads
    limited

  • Optimize cluster distribution
    for the auditors

GeoAnalytics

Geoanalytics is the use of location-based information (spatial or geographic nature) that are able to explain social, economic or environmental phenomena.

globo.jpg

Case: Deforestation Alerts Panel

desmatamento.png

Through an analytical panel with a strategic focus, the panorama of Degradation and Deforestation Alerts in the Legal Amazon region over the last few years is illustrated, supporting analysis and managerial decision-making work.

Case: Deforestation Polygon

poldesm02_edited_edited.png
polideesm01_edited_edited_edited.png

Through aggregation algorithms of deforestation areas, the solution is able to measure the speed of advance in each group of polygons and, with that, indicate which ones are active (ie, "live polygons"), making it possible to display them in an analytical solution to help the planning of environmental protection actions.

Chatbot Platform

Platform that provides cost and complexity reduction in the creation and lifecycle management of chatbots. With the use of Serprobots, the management of the development and infrastructure behind the software of a chatbot is abstracted. Everything is done visually in the platform interfaces. This makes it easier for stakeholders to configure and instantiate new chatbots quickly, making it easier to focus on what really matters, which is curating content and enriching a chatbot's artificial intelligence.

  • Consumption billing model

  • Visual interface for managing chatbots, without the need for code development

  • Abstraction and automation of operations and development (DevOps) tasks

  • Evolution and centralized maintenance of the technologies behind chatbots

chatbot01.jpg

Case: Serprobots 

serprobots01.png
serprobots02.jpg
serprobot incra.png

Ivo, Sol and Susi are examples of chatbot solutions developed by Serpro.

Natural Language Processing

It is a branch of Artificial Intelligence that aims to process and interpret human language, so that digital solutions have the ability to “understand” what is spoken and/or written.

Examples of applications:

  • Classification and grouping of texts

  • automatic summarization

  • sentiment analysis

  • speech recognition

  • information extraction

  • Identification of entities

GettyImages-507065943.jpg

Case: Recommendation System

Sgd.png

The gov.br Portal's recommendation system consists of improving the citizen's experience by offering digital services of interest, presenting options for services correlated with the citizen's search history.

Case: Approval of Medical Certificates

Case Homologação de Atestados Médicos.png

Problem:The approval of medical certificates of public servants is an expensive activity for the public administration, not to mention the time that is lost for the server to physically move to present the documents.

Solution:From the mobile application itself, through Artificial Intelligence, it was possible to perform the automatic approval of medical certificates.

Case: Standards in the Official Gazette of the Union

Case Padrões no Diario Oficial da União.png

Identification in the Federal Official Gazette of excerpts related to changes in rules relating to tax treatment for foreign trade.

Case: Composer of Petitions

Case Compositor de Petições.png

Petitioning assistant (elaborator of legal documents).

Problem:Repetitive activity that generates a large volume of work.

Solution:It provides initial petitions containing all the essential elements of the document, including the legal grounds that will be suggested according to the thesis.

Master Data Management (MDM)

Defines, unifies and manages all data that is common and essential for all areas of an organization.  This master data is typically managed in a single place or environment.

GettyImages-95769788.jpg

Master Data Management

mdm01.jpg

MDM project benefits

mdm02_edited.jpg

Other earnings

mdm03.jpg

Robot Process Automation (RPA)

Serpro's portfolio of digital solutions has increased with the arrival of RPA (Robotic Process Automation). RPA solutions operate on systems simulating a Human operator: speeding up the execution of well-defined and repetitive flows; minimizing operational errors; maximizing productivity and focusing on jobs that generate greater added value for the organization.

Through RPA the work sequence, which is performed by one person, can be roboticized. In this process, all navigation, clicks and commands are mapped and organized in the robot's execution flow, and can be executed according to demand and need. Which generates greater schedule management, speed and scalability.

GettyImages-911943966.jpg

benefits

  • Scalable (better dimensioning. Can add or remove robots as needed)   

  • High availability (maintenance only / scheduled downtime)

  • Optimization and standardization of Processes

  •  Repeats Elimination / Mitigation

  •   Decreased manual errors. Avoiding unnecessary rework and expenses;

  •  Non-Invasive Technology

  •  Support in compliance and auditing. Robots can work with 100% assertiveness. And RPA can help uncover non-compliant interactions.

  • Robots can support the work of numerous people simultaneously.

  • Optimize team productivity. Freeing up cognitive resources for higher value-added tasks.

where to use

  • Where to seek to reduce error rate

  • Complex flows 

  • High workload 

  • Operator deficit / High backlog 

  • Lack of personnel for new activities 

  • Difficulties / Limitations

  • Flows with rules and/or exceptions with lack of standards

  • Complexity in handling unstructured data. Where it can be complemented with the use of AI (NLP, ML, image/voice recognition)

rpa01.png

Robotic Process Discovery (RPD)

Uses Artificial Intelligence and Machine Learning to map and recommend processes for robotization via RPA technology.

The Discovery step occurs via monitoring the execution of the processes of interest (on the machines of the responsible users). In which data is collected for evaluation and recommendation by specialized algorithms. With subsequent acceptance and review by the business managers involved.

Identifies where there are bottlenecks in monitored processes and looks for opportunities to free up employees' cognitive time. Automating these manual and repetitive activities.

 

RPD benefits

- Find sequences or variations in processes that are not obvious to the naked eye;

- Identify the processes to be automated to maximize results;

- Take the guesswork out of identifying an automatable process, mapping how it is done, and designing the automation workflow.

rpd02_edited_edited_edited.jpg

Available services

rpd03.png

Analytics

It is a field that uses mathematical and statistical techniques to visualize data in an intelligent way, helping in decision making and generating knowledge for the business.