Technology Overview

Table of Contents
Executive Summary 3
Certified and robust technology components 4
Secure Passive Liveness Detection 4
First certified Passive Liveness 5
Liveness technology to minimize fraud 5
ROC Curves 6
Certified Liveness based on ISO/IEC Standard 7
World-class, proprietary Face Recognition 8
User feedback and automated photo capturing 10
General facial recognition model features 12
General performance features 14
ROC Curves 16
Robust ID Document Validation 16
Image & Data Capture 17
Face recognition 17
Document verification 17
ROC Curves 18

Flexible integration 20
Engines 20
Identity Servers 20
SDKs and APIs 20
Implementation and solution architecture 22
Security Framework 24
Application security and Web message transmission 24

Incode’s key differentiators 25

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 2

AUGUST 2022
Executive Summary
Incode has developed world-class proprietary technology that enables validating someone's physical
identity and moving it into the digital world. Two steps are needed to achieve this: 1.Facial Authentication
(comprised of Facial Recognition and Liveness Detection), and 2. IdentityDocument Validation.

Incode’s technology is recognized as one of the most secure Identity solutions that exist:

• Our facial recognition technology is one of the best in the world, as evaluated by the
National Institute of Standards and Technology (NIST) in its July 2019, 1:1 FRVT report,with
the best combination of facial recognition speed and accuracy.
• Incode is the first company to have received a certification for our passive liveness
technology following the ISO/IEC 30107-3 Presentation Attack Detection standard, as
certified by iBeta following the NIST guidance.
• Incode has also developed proprietary technology to perform tests on the ID
documents presented and detect potential fakes, tampered, or photoshopped
documents.

Incode’s Omnichannel Identity Solution gives Audi a full world-class solution; fulfilling high standards
and expectations. The following is a summarized diagram to overview the fullcomponents to be able
to achieve great experiences.

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 3

AUGUST 2022
Certified and robust technology components
Incode Omni solution utilizes three in-house models for enabling the solution, the models areas
follows:
1. Liveness Detection
2. Facial Recognition
3. ID Validation

All the three models are based on Deep Learning. Deep learning (also known as deep
structured learning or hierarchical learning) is part of a broader family of machine learningmethods
based on artificial neural networks. Learning of these models varies can besupervised, semi-
supervised or unsupervised, depending on the model.

Secure Passive Liveness Detection

Incode is the first company in the world to have been certified following NIST and ISO/IEC-3017
standards for passive liveness detection, meaning no action is required from the user.

Incode delivers world-class proprietary technology. Incode’s technology is recognized as one of the
most secure Identity solutions on the market today.
● Our facial recognition technology is certified in passive liveness - endorsed by iBeta
ISO/IEC-30107-3 [certified L-1]
● Incode technology is the first certified technology in the world under the U.S.
Department of Commerce's NIST standards (2019, 1:1 FRVT report); with the best
combination of facial recognition speed and accuracy (faster than the competition)
● The Guidance in Digital ID issued by the Financial Action Task Force primarily refers to
NIST digital ID guidelines to present strategies for mitigating threats to identity proofing
and enrolment processes

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 4

AUGUST 2022
First certified Passive Liveness
Avoid embarrassing moments to the users, they won’t need to blink or make strange facial
expressions.

Additionally, provide a secure Identity landscape. Incode is the only technology in the world to
meet US Government standards. Incode’s liveness technology passes NIST certification
requirements for liveness detection using a passive algorithm that requires no action from the user
(Level one certification from iBeta Quality Assurance accredited by NIST/NVLAP [NVLAP Lab Code:
200962] to test and provide results to this Presentation Attack Detection - PAD testing standard).

Liveness technology to minimize fraud

Reduce frauds, catch if someone tries to use a video or photo instead of an activity attempted by
the real person.

Incode’s liveness detection is what many providers in the industry refer to as “Level 5”, which is the
most sophisticated one. In essence, it models the light on the face, estimates depth and generates
a 3D model of the face. This way it’s very difficult to spoof it with photos, videos and other
elements, as it has a very high true positive vs false positive rate.

Its main features are:

● High true positive (accuracy) vs false positive (mistakes) rate (TDR vs FDR)
● It doesn’t need any user feedback, movement, blinking etc; works passively
● It can work with one or more frames
● It’s fast (Less than a second)
● Can run on server and on-device
● It’s not hardware-based, runs on commercial devices (from low-end smartphones)

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 5

AUGUST 2022
ROC Curves
Thresholds are usually chosen based on the ROC curves. In our particular case, for all our models
we choose thresholds at the points where FPR=0 which means that the model doesn't have any
tolerance to false positives and then shift them by approximately 5% to make sure they will work
for new data that the model hasn't seen before, for example, if the threshold is 0.5 we will put 0.525
as the threshold.

To make a system more secure (like in case of liveness) the threshold can be shifted even higher if
the higher false negative rate is not an issue. We highly don’t recommend moving thresholds
without approval from Incode’s side because TPR/FPR dependency isn’t linear (ROC curves show
that) and their values highly depend on the input data. Also, financial entities shouldn’t consider
the threshold as a probability so for example the threshold 0.3 for recognition model doesn’t mean
the probability of successful recognition equals 0.3. The threshold is just some number that helps
to classify input data.

This is the ROC curve graph that shows the TPR (accuracy) vs the FPR (mistakes):

Note: The most recent version of ROC will depend on the R&D of the technology

Threshold
Low severity level: TPR = 0.984234 : FPR = 0
Medium severity level: TPR = 0.966216 : FPR = 0
High severity level: TPR = 0.954955 : FPR = 0

For liveness, we collected our own dataset because none of the public datasets are representative
enough for real-life applications.

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 6

AUGUST 2022
Accuracy of both FR and liveness models is above 0.999. However, we don’t use accuracy as a
metric for our FR and liveness models because we don’t allow any tolerance to false positives for
them.

Certified Liveness based on ISO/IEC Standard

ISO/IEC 30107 - Liveness Testing Global Standard. Is the International Organization for
Standardization’s (ISO) testing guidance for evaluation of Anti-Spoofing technology, a.k.a.,
Presentation Attack Detection (PAD). “bio” “metrics” literally means to measure live human
physical characteristics. Ironically, it took until late 2017 for anyone to release official guidance on
how to determine if the subject of a biometric scan is actually alive.

Incode liveness - high level explanation

The following image shows how our liveness works at a high level

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 7

AUGUST 2022
Model testing and certification
The test method was designed to simulate user enrollment into a biometric authentication system.
This test did not perform matching and was purely a test of liveness detection effectiveness.
Testing was conducted in accordance with the contract for a level of spoofing technique that only
utilized simple, readily available methods to create artefacts of a genuine biometric for use in the
presentation attack. The subjects for the test effort were cooperative meaning that they were
willing and able to provide any and all biometric samples, including high quality photos and videos
of their likeness. The test time for each PAD test per subject was limited to eight hours. This is
considered a Level 1 PAD test effort.

The test method was to apply one bona fide subject presentation that alternated with 3
presentations of each species resulting in 300 Presentation Attacks (PAs) and 100 bona fide
presentations per artefact. The application displayed messages with a green lock for successful
liveness confirmation oar red lock for an unsuccessful liveness conformation.
On the smartphones used in the test, iBeta was not able to gain unauthorized access (simulated
enrollment) with a presentation attack 300 times with each of 5 species of attacks. With
300attempts for each species, the total number of attacks were 1,500 and the Presentation
Attack (PA) success rate was 0%.

Click here to review Incode’s Liveness certificate by iBETA >>

Other features
• Depth-based liveness detection modeling
o Utilizes presentation attack detection based on depth map classification using
trained neural network
o Does not need any type of feedback from user
• Liveness check performed in 200ms on server
• 256 x 256 minimum face size
• Outputs confidence level so that the Client can incorporate it to its risk algorithms

World-class, proprietary Face Recognition

The following are some of the general face recognition model features that Incode offers:
● Proprietary deep-learning face recognition model and product developed by Incode
Technologies Inc.
o World-class, face recognition technology and a leader in enterprise-grade facial
recognition recognized by the United States Department of Commerce’s National
Institute of Standards and Technology (NIST)

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 8

AUGUST 2022
 o Incode’s huge, diverse photo database provides data to power its best-in-class
machine learning model
o Expertise in low-light, in-the-wild photos
● Log that records all actions executed by face recognition application (with errors and
exceptions) that can be reviewed in any text editor
● Runs independently (on device) with no internet or connected to server
● Able to run on photos with 64 pixels (24 bits per pixel) between the eyes
● Biometric facial templates created and stored from faces (mathematical representation)
- Cannot be reverse-engineered to generate face images from templates

NIST’s results have positioned Incode at the top of the list among enterprise-grade facial
recognition providers in terms of accuracy as well as template creation speed as seen in the chart
below as of July 31st, 2019 evaluation:

Click here to review NIST’s face recognition rank report >> using our production-based model

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 9

AUGUST 2022
User feedback and automated photo capturing
Incode’s system automatically takes a photo of the user.

Incode’s system gives the user feedback to ensure the optimal photo is taken.

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 10

AUGUST 2022
Incode’s system performs multi-factor liveness detection to ensure the user is a real person. The
system does not allow bad actors to log-in.

Incode’s system compares the selfie with the photo from the accepted ID to verify the person’s
identity. The Client can modify these thresholds and set logic to pass, fail or require applying
individuals to visit a branch office for additional verification.

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 11

AUGUST 2022
With an Incode-led integration, existing user can log back in to continue or finish applications.
Incode enables the client to re-engage with customers to reduce churn and drive adoption.

General facial recognition model features

The following are some of the general face recognition model features that Incode offers:
• Proprietary deep-learning face recognition model and product developed by Incode
Technologies Inc.
o World-class, face recognition technology rated as top 20 on server and top 3 on phone
by the United States Department of Commerce’s National Institute of Standards and
Technology (NIST)

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 12

AUGUST 2022
 o Incode’s huge, diverse photo database provides data to power its best-in-class
machine learning model
o Expertise in low-light, in-the-wild photos
• Log that records all actions executed by face recognition application (with errors and
exceptions) that can be reviewed in any text editor
• Runs independently (on device) with no internet or connected to server
• Able to run at on photos with 64 pixels (24 bits per pixel) between the eyes
• Biometric facial templates created and stored from faces (mathematical representation) -
Cannot be reverse-engineered to generate face images from templates
• Processing available online in real-time or via batch processing, depending on the Client’s
requirements
• Photos captured with Incode’s products are stored with a timestamp in image and/or photo
metadata
o Photos stored at 200 pixels per inch (PPI)
• Outputs confidence level so that the Client can incorporate it to its risk algorithms
• Support for human verification to improve model

Capture features
• Photos are taken exclusively in real-time through digital platform (mobile web, desktop
web, native iOS and native Android)
• Faces in photos are framed to indicate adequate zoom and resolution
• Feedback given to user with needed feedback (e.g., “move closer”)
• Photos automatically captured when face is correctly in frame
• Photos can also be captured with manual press to capture button
• Creation and management of templates (mathematic representation)

Selfie comparison with ID photo features

• Model compares selfie and ID photo and returns confidence level of same person
• Outputs confidence level so that the Client can incorporate it to its risk algorithms

Log-back-in features
• Users are automatically enrolled in a face recognition system
• Because Incode performs “true face recognition” (not just selfie comparison), enrolled users
can log back into their application at any time to resume an application process or check on
the status of an application

Metrics and configurability features

• The Client can view and set thresholds (e.g., accepted face recognition confidence, selfie v. ID
photo matching confidence, liveness detection) using Incode’s flexible, user- friendly
thresholds UI
• The Client can create dashboards to select and monitor KPIs (e.g., average face recognition
confidence, time to create and store template, etc.) using Incode’s flexible, user-friendly
dashboard UI

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 13

AUGUST 2022
 • All metrics are outputted so the Client can incorporate into its risk algorithms and customize
decision logic

General performance features

• Scalable microservice architecture with load balancers in front
• Processing and response time of Incode’s facial recognition and liveness detection API is less
than 1 second

Facial recognition performance features

• Template creation under 100ms (using Incode’s services /recommended infrastructure)
• Support for 1:1 face recognition
• Face template size: 1 KB
• Support of low-resolution photos: minimum 64 pixels between eyes
• Distributed architecture can support more than 1,000,000 concurrent users
• Maximum server template storage: Unlimited
• Support for 1: N face recognition
o Verification process to review if there are other individuals with the same photo
and different ID
o Can be used to ensure user’s face being enrolled does not correspond to that of a
Client employee
o Can be used for de-duplicating existing customer/employee databases
o With 100,000 enrolled templates, recognition takes 2 ms.
o With 100,000,000 templates on server, recognition takes 100 ms.
• Already supports future face recognition uses cases such as:
o Long distance face recognition (someone entering a Client branch)
o ATM login
o Transaction verification on branches and online
o Queue management systems
• Multiplatform:
o Desktop and Mobile Web (Safari’s latest release version, Chrome, Firefox, Edge,
Electron, Internet Explorer, Opera) SDKs
▪ Support HTML wrapper in iOS and Android native apps
▪ Support for fully-web based experience
o Native iOS SDK
o Native Android SDK
• Can run all from server
o Supports GPU and/or CPU face recognition processing
• Can run all from Client (in iOS and Android)
o Can run with or without Internet connection if using native application
• Can be fully hosted on premise
• Can be fully hosted on Client’s private cloud

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 14

AUGUST 2022
• Can be hosted with database (facial templates/PII) on premise and all other processing
performed in secure, auto-scaling Incode’s AWS cloud
• Complies with:
o FIDO
o ISO/IEC 19795-1
o ISO/IEC 19794

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 15

AUGUST 2022
ROC Curves

True Positive – matched faces for one person, False Positive – matched faces for different
person

Threshold = 0.3: TPR = 0.997179: FPR = 0

True Positive – matched faces for one person, False Positive – matched faces for different
person Threshold = 0.6: TPR = 0.942718: FPR = 0

Robust ID Document Validation

In the case of an ID Document, Incode has developed its own ID Validation Tests in order to
verify the ID Document is a real and valid one. This Validation score is derived from different
analysis applied like the following ones with proprietary Incode Algorithms.
At a high level, the Incode validation process considers the following activities, which enable
performing tests to determine the authenticity of the document.

Image & Data Capture

Incode SDKs are capable of automatically detecting the ID document in the frame, and then
checking if the image quality enables proper document verification.
• Document detection. Recognize document borders and calculate best possible
image capture instantly.
• Light and surface analysis. Runs a series of the technical checks to evaluate quality
and conditions of environment.
• Image quality assurance. Check if image quality enables proper document
verification.

Face recognition
To verify that the holder of the ID Document is the legitimate user.
• Face matching. ID verification performs a face matching between document portrait
and selfie photo.

Document verification
• Document classification. Based on the document composition and patterns, it
recognizes the type of the document and the sections of data available.
• Document verification. Performs an extensive series of security checks, to assess
its authenticity.
• Data scanning. Once ID is verified extracts the personal and biometric information
to perform more tests.
• Data validation. More tests are executed to check data tampering and document
• validity like expiration date and crosschecks.
• Incode also offers the option of verifying the driver’s license information against the
official databases of 37 State DMVs.

During our tests, Incode’s Machine Learning team sampled the scale of scores (0-100) and
calculated the accuracy for each possible threshold to come up with the one (65) that gives
the best result.

Major Incode Checks

• Incode FakeCheck: Technology to identify and reject fake documents. ML model that
can detect patterns that differs from the original ones. For example: slight difference
in layout of the ID, different font or font size.

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 17

AUGUST 2022
 • Incode TamperCheck: Technology to identify IDs that have been physically or
digitally tampered to steal identity. ML model that can detect if face photo is modified
or if two face photos on ID are not the same.
• Incode ID Liveness: Technology to validate if the ID captured was a physical one vs
printed or screen-based capture. ML model that analyses patterns that could be
evidences of capturing from some type of screen (computer, tv, mobile phone).

ROC Curves
ID liveness detection for Screen detection

True Positive – live id is live, False Positive – spoof id is live

Threshold = 0.5: TPR = 0.984430; FPR 0.045699

ID liveness detection for Paper detection

True Positive – live id is live, False Positive – spoof id is live

Threshold = 0.5: TPR = 0.984430; FPR = 0.100000

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 18

AUGUST 2022
ID retouch detection (tag 0.2.0)

True Positive – original id is original, False Positive – retouched id is original

Threshold = 0.5: TPR = 0.998848; FPR 0.003556

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 19

AUGUST 2022
Flexible integration
Incode Omni is completely flexible and supports different integration architectures. Our
platform follows a Microservices Architecture. The microservices are connected to one
instance database; we follow a replication scheme following active/active scenarios to
provide redundancy and high-availability. And the global application can be connected to
other services such as queue management, video-conference communication service, other
services connected via an API, etc.

Engines
Incode’s “identity engines” are Incode’s core AI models. Most of Incode’s models are based on
machine-learning techniques. Incode’s “ID verification” model, for example, considers many
features of an ID and then uses computer vision and learning to determine if the ID is real or
not. These powerful models allow Incode to perform fingerprint recognition, face recognition,
ID verification, and other identity-based use cases.

Identity Servers
Incode also maintains several identity servers which help us develop a customer’s full digital
identity. These services, for example, scrape data from online accounts from utilities, other
banks or social media sites, run queue management software for Clients, and help send
money amongst customers around the world.

SDKs and APIs

Incode’s Omni Platform offers prebuilt product flows via ready to integrate SDKs. However,
our customers have the flexibility to develop their products and use cases using either our
SDKs or our underlying engine APIs. This flexibility allows our customers to customize their
products and experiences as per their needs.

Incode Omni’s engines and identity servers are not, however, enough to build great customer
experiences.

Let’s assume, for example, that Company wants to check if a user’s face is already registered
in the database. Incode does use the underlying “1-to-N face recognition engine” model to
search for the user, but this is not sufficient. Incode’s supporting SDKs are needed to help the
user capture their picture on the phone, apply brightness and glare modeling to the photo,
and give the user necessary feedback as they are capturing the selfie.

If Company is building this experience within an Android app, the Client would integrate with
Incode’s “Face Login Native SDK” for Android (native) for this use case. The Client can directly
integrate with this SDK. The Client does not need to integrate with both the relevant
CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 20
AUGUST 2022
supporting SDK and the relevant engines. Incode ensures that all supporting SDKs come
pre-integrated with our engines. This reduces the integration complexity and increases
the time to market for the application.

Develop products and use cases using our low-level engine APIs, or develop products using
our experience-ready plug and play SDKs or via Incode’s Admin portal self-service flows.

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 21

AUGUST 2022
Implementation and solution architecture
Incode solutions can be deployed on-premises, SaaS or Cloud Managed Services.
Incode Omni is completely flexible and supports different integration architectures. Our
platform follows a Microservices Architecture (MSA) on Container Platforms (Docker or
similar). We use a software container platform to provide isolation, portability and scalability.
In addition, it allows for greater integration capacity.

The following diagram shows the architecture scheme with Incode’s main micro-services,
main instance database and additional modules.

Leveraging Container Platform for micro-services provides the following benefits:

• Isolation. Each application can only access its own container space. If an application
gets compromised, it won’t be able to access other applications.
• Security by default. Swarm provides features that are enabled by default that
improve application security such as read only containers.
• Portability. A container could be run anywhere since the whole definition stays
within.
• Scalability. Applications can be easily scaled by launching more containers (with
orchestration).
• Easier elasticity. Dynamic provisioning becomes straightforward.
• Images of all microservices are provided through docker hub
• Documentation for deployment and management of services
• Regular updates of images with new features
• Regular security updates
• All releases ensure backward compatibility

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 22

AUGUST 2022
To run the Incode platform, Client can go with any of the following infrastructure options:

1. Incode Cloud (AWS)

Use Incode Cloud on AWS Infrastructure; for the Incode Cloud (AWS) solutions, Incode can
provide dedicated and managed private cloud services to the customer if required

2. Hybrid model: Incode Cloud (AWS) + Persistent data On Premises

Using Incode Cloud infrastructure but all data is transferred and persisted at
onpremises infrastructure

3. Client’s private cloud

Using a Container Platform (Docker or similar), components and services are
deployedon container images

4. On-premises
Using a Container Platform (Docker or similar), components and services are
deployedon container images

Incode's 3 security principles: confidentiality, integrity and availability.

The microservices are connected to one instance database; we follow a replication scheme
following active/active scenarios to provide redundancy and high-availability. And the global
application can be connected to other services such as queue management, video-conference
communication service, other services connected via an API, etc.

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 23

AUGUST 2022
Security Framework
Incode incorporates robust security measures for a cloud environment, we follow our Security
Framework, which considers:
• Web Application Firewall
• IP whitelisting for API gateway
• HTTPS (TLS 1.1., TLS 1.2) to encrypt data in transit
• API access keys
• API Message Protection with JWT
• File system-based encryption for data at rest (AES-256)
• Database encryption (AES-256)

Application security and Web message transmission

Incode bases its applications with security as the main premise:
• HTTPS for communication between applications and Incode micro-services
• JWT: all sensitive data transferred between Client and server is encrypted within JWT (Json
Web Token)
• Access controls and roles
• API access keys for additional protection
• Restricted HTTP methods: certain requests will be available to administrators with token
• Input validation: applications validate the type of content and requests
• Error handling is carried out at the application level with Global handlers

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 24

AUGUST 2022
Incode’s key differentiators
The following are some of the compelling reasons that differentiate Incode over other
vendors.
• Incode Omni is the only omnichannel biometric solution in the Western world*
• Incode offers pre-built End to End products, this means shorter time to market for
your solution - our platform is completely modular and flexible.
• Incode focuses on finding and solving the edge cases of real-life biometrics – works in
real life and with all devices and platforms
• We have the only passive liveness test in the world that is certified by NIST
• Best enterprise facial recognition according to NIST 1:1 report on July 30, 2019
• We are obsessed with improving the user experience
• Significant traction with global organizations
• Incode offers an all you can eat subscription model that is unique in the industry
• Our lightweight SDKs are easy to integrate and implement
• We have a modern microservices based architecture
• Our solution can run-on low-end devices as well as on the server*
• Our solution runs on-premise, in the cloud, or in hybrid mode*

* Note: See “Incode – Product Overview” for more information

ABOUT INCODE
Incode is a provider of secure biometric products for banking, payment and retail industries. The
company’s flagship suite, Incode Omni, is an end-to-end omnichannel identity platform that
enables seamless access across multiple channels to reach and engage the next-generation
consumer.

Recognized as the leader in enterprise-grade facial recognition technology by the US

Government (NIST),Incode Omni is already used by some of the world’s largest banks, financial
institutions, governments and retailers. Incode is based in San Francisco with offices in Europe
and Latin America.

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 25

AUGUST 2022
 Incode Technologies, Inc.
221 Main St, Suite 520
San Francisco, CA 94105
www.incode.com

CONFIDENTIAL INFORMATION PROPERTY OF INCODE TECHNOLOGIES INC. 26

AUGUST 2022