The world of finance and accounting is experiencing a transformational shift that is changing the way we manage money, invest, and work. As we gather here today, we are witnessing the emergence of a new era in finance that is powered by automation and innovation.

Having a finance and accounting background, I have personally witnessed the challenges that come with the adoption of automation in the industry. I saw technology products that were not complete on their own to provide total solutions, automated methods that fell short in the face of the adaptive situation, and an access rule that compelled enterprises to compromise on collaboration and data security in unacceptable ways. We saw a demand for an AI technology that would require a different company to develop. That's how Xaana came to be.

As an AI company founder and applied developer, I get asked this often so I might as well begin with this question,” Will machine intelligence replace human intelligence and ultimately human workers in the near future?”.

In an economy where data is changing how companies create value — and compete — experts predict that using artificial intelligence (AI) at a larger scale will add as much as $15.7 trillion to the global economy by 2030. The vision of the future of work has taken the shape of a zero-sum game, in which there can only be one winner.

Historically, the phrase “artificial intelligence” was coined in the late 1950s to focus on the high-level or cognitive capability of humans to reason and to think. Seventy years later, however, high-level reasoning and thought still remains elusive. The past two decades have seen major progress—in industry and academia—a search engine or a chatbot can be viewed as an example, with natural language processing, which supplements the ability of a human to write and communicate. While services of this kind could conceivably involve high-level reasoning and thought, currently they sadly don’t; they mostly perform various kinds of string-matching and numerical operations that capture patterns. Being in this space, the sad thing about artificial intelligence is that it lacks artifice and therefore intelligence.

Even to this day, there does not exist a univocal and shared definition of the term “Artificial Intelligence” because it is a concept that includes a very large number of related topics to different disciplines, from neurology to computer science, from neurobiology to neurophysiology (and in general all disciplines that study the human brain) to math and so on.

It has been my life’s work to understand the gaps in AI and come up with a Hybrid Intelligence system that combines the very different processing strengths of Artificial Intelligence and the human brain to result in a symbiotic Augmented Intelligence platform. The platform has already allowed decision makers across industries to spend less time setting up the digital chess board and more time analysing and planning the right moves to win every operation. We call this Augmented Intelligence at Xaana. To answer the question, this is the present of intelligent work. Augmented Intelligence is much greater than the sum of its parts, and it comes down to you to either embrace it or resist it because it's here. It represents a symbiotic relationship between a man and machine. Rather than replacing us, augmented intelligence helps our decision-making ability—and by extension, our intelligence too.

In our current dynamic capital market, the platform demonstrates a fast and accurate machine learning capability to handle complex data and has gradually acquired the potential to become a "financial brain". It specialises in systems integration and interoperability solutions, with a focus on knitting disparate databases, IOT systems, apps, and platforms together, so each component works together harmoniously, producing insights in real-time and delivering a capability greater than just the sum of its elements. For example, our technology, when provided with big data, can integrate long-tail markets and mitigate information asymmetry to improve the efficiency of fund allocation and financial risk management. Many financial applications, such as investment, lending, credit, security, insurance, and customer service, are inseparable. There’s a gap and need to combine the structural financial report data and unstructured behavioural data of financial users via a knowledge graph which we call the multi-layered ontology. Ontology assimilates data and models into a comprehensive semantic foundation layer and provides an “operational picture” throughout all parts of the enterprise.

End-to-End Financial Solutions: Ensure every cent provides tangible impact.

Xaana’s Platform unifies financial data across silos, systems, and functions without a full system migration — Creating a collaborative, secure foundation for financial management professionals to analyse portfolios, track budgets, and intelligently reallocate funds. Once in place, an integrated foundation can support a wide range of financial activities, from budgeting to auditing and all in between. Budget analysts are frequently forced to depend on static financial statistics, which limits their capacity to manage risk and adapt budgets when priorities shift. It enables analysts to proactively spot deviation and intelligently reprogram money, all while communicating smoothly with internal and external stakeholders. Detecting inconsistencies as billions of transactions flow through current financial systems, identifying mismatched transactions and unauthorised payments and sending warnings to analysts for corrective action.

Financial fraud is difficult to identify and prevent, with digital transformation speeding up sophisticated cyber attacks, and new unregulated digital assets like cryptocurrency increasing exposure. Xaana’s Platform combines multi-layered security with an industry-leading architecture to provide solutions for: Transaction Monitoring, Investigations, KYC, Customer and Transaction Risk Rating, and Enhanced & Customer Due Diligence

1. Fraud detection on your own terms

Companies that do not have machine learning experts, can use AutoML to add ML-based fraud detection capabilities to their business applications in minutes while companies with a dedicated team of data scientists can bring their models or build models on CustomML to develop highly specialised fraud detection solutions in days.

2. Prevent and detect online fraud in real-time

Our Fraud Detection ML solutions scores the risk of an event in real-time, allowing customers to instantly apply containment or remediation measures designed to block or deny fraudsters and fast-track low-risk activity to provide better customer experiences for legitimate customers.

3. Give fraud teams more control

By automatically handling the complex tasks required to train, tune, and deploy a fraud detection model, our Fraud Detection ML Solutions make it possible for users who aren’t machine learning experts, but are familiar with fraud issues, to participate in developing and updating highly accurate models.

4. Powerful AI/ML Foundation

Our ML-based entity resolution and powerful network-based risk models make linking consumers, networks, and unknown counterparties simple and accurate. This provides the bank with a stable, future-proof foundation along with a central client file that can be dynamically updated across functions.

5. Risk-Based Models & Scenarios

The platform takes a hybrid approach, giving full out-of-the-box modules, scenarios, and ML models while also providing the flexibility required to fulfil regulatory standards dictated by your specific client base and risk tolerance. This enhances alert accuracy and efficiency while providing regulators with more high-quality, automated SARs and reports.

6. Trusted Security

Xaana’s Platform is known for its unrivalled granular security measures. Encrypts all data in transit and at rest, and it provides high auditability and role-based access control frameworks so that teams may access and share data without compromising security.

7. Best-in-Class Tooling & Reporting

Analysts can swiftly triage warnings and ask follow-up questions to uncover the entire network of associated companies with a holistic view of transaction, relationship, counterparty, KYC, and sanctions risk. Built-in workflow management allows for prompt and appropriate escalation, triage, and review — with complete cooperation and transparency across all relevant departments.

Xaana creates needed reporting information with full data lineage to source systems after an enquiry is completed, dynamically updates reports as information changes, and informs analysts of new risks.

8. Commercial Off-the-Shelf Product

Xaana’s Platform serves as the foundation for the Next Generation Financial Crime Solution as a productized, out-of-the-box solution. It comes with a dependable, scalable backend designed for complex data environments.

9. Speed to Implementation

The proposed Solution can be rapidly implemented and will deliver a significant portion of the solution functionality out-of-the-box - meeting needs on an expedited timeline of weeks rather than years.

10. Foundation for Future Capabilities

As the world changes, the data infrastructure is designed to adapt and scale along with changing organisational needs.

11. Cost-Effectiveness

It is a fully integrated SaaS solution with a fast time to value and a plethora of capabilities to reduce Total Cost of Ownership – even when implemented on-premises. Organisations save money on labour, maintenance, support, and operating costs by bundling a suite of flexible applications.

12. Openness & Interoperability

It is an open, extensible, and scalable system. All data in the platform can be securely exported in industry standard, open formats for use in other systems, and all of the logic used to integrate our clients’ data is available and usable in any environment.

Use cases

1. Payment or transaction fraud detection

The event of interest is an attempt to complete an online purchase or make or process a payment online. One common example in the e-commerce space relates to a “guest checkout”. The transaction involves a user who does not have account history or has selected a “guest” checkout option for a more anonymous experience.

2. New account fraud

The event of interest is the act of signing up or registering for a new account. Fraud starts when a bad actor creates fake, stolen, synthetic identities, or generates multiple accounts, often through the use of bots. Once identity is established on a digital platform, executing an attack is easier.

3. Account takeover

The event of interest is a login attempt for a legitimate user account. Account Takeover refers to the situation where a legitimate user’s login has been compromised, either because a bad actor has stolen their user id and password, purchased them on the dark web, or managed to guess it.

4. Promotion abuse

The event of interest is typically the act of a user redeeming a benefit granted via a demand generation or marketing promotion. Bad actors will access a legitimate user’s account and drain loyalty credits or points via transfer or purchase. They will also create multiple fake accounts to exploit promotions such as a free trial or free credits that come with a new account, or perform a self-referral to get a referral bonus.

5. Authentication

During online account registration, machine learning-powered facial biometrics can enable identity verification for any situation. With pre-trained facial recognition and embedded analysis capabilities, you can add this to enhance your user onboarding and authentication workflow with no machine learning expertise required.

Documents come in various file types, varied formats, and contain valuable information. In most cases, you are manually processing the documents which is time consuming, prone to error, and costly. Not only do you want this information quickly but likely need to use the information within those documents for downstream applications. To help overcome these challenges, Xaana’s Machine Learning (ML) now provides you choices when it comes to extracting information from complex content in any document format such as insurance claims, mortgages, healthcare claims, contracts, and legal contracts.  

Benefits

1. Higher accuracy of data

Using ML can help you process documents faster and more accurately, reducing errors caused from manual entry. In cases where data needs to be 100% accurate to empower humans to review processes.

2. Faster data processing

Implementing intelligent document processing can help you accomplish weeks or months of work in a matter of days.

3. Improved employee productivity

Machine learning removes the manual process of pulling out insights from documents and entering information into various systems, enabling your employees to spend more time on value-adding business tasks.

4. Cost savings

Automating document workflows reduces the complexity of data extraction and analysis, reducing the average cost per document.

Industry

1. Banking

Mortgage packets come with varying document types such as tax filings, W-2’s, paystubs, and applications which oftentimes need to be split and classified. Using Xaana’s ML you can extract the most important information out of mortgage applications such as asset valuation, credit score or property value using a combination of Intelligent Optical Character Recognition (iOCR) and Natural Language Processing (NLP) to speed up response times to your customers.

2. Insurance

Many insurance forms have varied layouts and formats which makes text extraction difficult. Using machine learning, you can extract relevant fields such as estimates for repairs, property address or case ID from sections of a document or classify documents with ease. By combining text extraction and NLP, you can process insurance forms such as insurance quotes, binders, ACORD forms, and claims forms faster, with higher accuracy.

3. Capital Markets

Financial proxy statements, SEC filings (10-K, 8-K, 14A, 497K, etc.), KYC forms, tax documents and more come in dense text format or mixed with tables and text making it difficult to process using traditional methods. Using Xaana’s ML, you can process various formats and file types using iOCR and NLP combined to extract table formats and derive entities from documents and use custom models to recognize the entities and classify documents.

4. Manufacturing

Bill of materials (BOM), Certificates of Analysis (COA), and Purchase Orders (PO) are a major part of a manufacturing operation, which today is usually manual and time consuming. Using Xaana AI, you can now automate the process by extracting text from contracts, identifying specific fields and values, and using the data to inform downstream systems in your manufacturing systems.

Cross-Industry

1. Legal

Processing documents, such as agreements, court filings, or legal dockets, is a difficult task for legal teams. Contractual documents are often in non-standardized formats. The typical workflow for reviewing legal filings involves loading, reading, and extracting case number, parties involved or legal entities from the documents, requiring hours of manual effort. Using iOCR and NLP to extract text and specific terms can automate this process, with higher accuracy.

2. Accounts Payable

Invoices and receipts are vital to all organisations and many times those types of documents come in various layouts. Using Xaana ML you can automatically extract valuable information within those documents to automate your business, reduce cost per page and manual effort.  

Financial markets involve assets, traders, online platforms, supply chains, and logistics, and they can be considered as multi-agent systems (MASs). Learning to play a multi-agent game is crucial in robust credit systems, supply chain management, dynamic asset pricing, and trading mechanism design. To this end, three challenges should be addressed. First, given scalability problems, a small

number of agents is usually considered, which hinders the deployment in businesses that involve large-scale agents. Second, game-theoretic-based agents rely on the assumption of perfect rationality of individuals, which is difficult to apply in real-world scenarios. Lastly, cooperative actions and rewards must be quantified. Moreover, a profound discussion has been conducted on the future of economic AI, wherein AI systems engage in business transactions with other AI

systems as well as with firms and people. Therefore, Xaana provides a comprehensive design to capture the rules of interaction to elicit truthful reports and promote fair contributions. We can associate our platform with a reputation to prevent the problems of moral hazard and adverse selection and further promote cooperativeness when completing financial transactions.

A key component of financial agents is the ability to provide explanations for their decisions, predictions, or recommendations. However, current AI systems usually make black-box predictions but rarely explain the process of decision making in a way that is meaningful to humans. We stress that explainable decisions are important to financial users. For example, investors require understanding before committing to decisions with inherent risks. In general, explainable AI should be able to identify the properties of the input (i.e., feature interactions) that are responsible for the particular decision and should be able to further capture the causal relations to answer counterfactual questions. The key to designing our platform is providing causal inference to better understand the real-world environment and support interactive diagnostic analysis that faithfully replays a prediction against past perturbed inputs to measure feature importance.

The impact of fraud and economic crime on all organisations worldwide still

reaches high-record levels. The most frequently committed fraud scheme is asset misappropriation. Laws, regulations and forensic methodologies cannot efficiently cope with the growth pace of novel technologies, which translates into late adoption of measures and legal voids, providing a fruitful landscape for malicious actors. In this regard, the features offered by Xaana’s blockchain technology, such as immutability, verifiability, and authentication, enhance the robustness of financial forensics.  Data acquisition is one of the most critical steps during forensic investigations. This is interpreted as maintaining a chain of custody for data and performing data integrity validation to ensure tamper-proofness. The importance of chain of custody is highlighted when the fraud investigator has to compile financial information for decision-makers (senior management, prosecutor, etc.). Blockchain offers different features, such as auditability, security, decentralisation and transparency. Blockchain applications in the financial context are already a fact.

The amount of cyber challenges faced by the financial sector requires cooperation between different entities and actors both at national and international

levels. According to the Basel Committee on Banking Supervision, there are different kinds of fraud: internal/occupational frauds or external frauds. We provide an implementation based on Ethereum and smart contracts to preserve the chain of custody as well as the trail of events. Overall, our solution enables various features and benefits, such as integrity verification, tamper-proof, and future enhancement of similar investigations. Therefore, it aims to close the gap between the technicalities of financial investigations and legal procedures, enhancing the robustness of the current state of practice.

To detect deviations and specific behaviour in documents and social networks, fraud investigators can use techniques of sentiment analysis augmented with graph analysis to discover the degree of connectivity between individuals in Xaana’s ML Platform. An essential aspect of visual analysis is the interaction of systems, letting investigators select, explore, and filter the visualised data. One of the main approaches to using Xaana’s machine learning techniques is to classify data according to some criteria (e.g., identify multiple transactions at a customers’ account in quick succession) and to predict the occurrence of fraud through the use of neural networks, Bayesian learning, decision trees and association rules. Other approaches also include rule-based expert systems and process mining models which can extract and detect further data patterns. Due to the modus operandi of the malicious actors, the use of Xaana’s natural language processing detects authorship of documents, including the use of specific keywords to tag specific transactions or operations, which can help in the detection of embezzlement. Also, capture and analyse multimedia forensic including processing and storing CCTV video evidence, and image-based provenance and integrity.

Integrity: The events data as well as other evidence are not altered or corrupted during the data transfer and analysis due to the use of hashes.

Traceability: The events as well as other evidence can be traced from their creation till their destruction since every interaction is stored in an immutable ledger.

Authentication: All the actors and entities are unique and provide an irrefutable proof of identity due to the use of asymmetric cryptography.

Non-repudiation: Each action can be related with its author in a graphical network, enabling strong accountability guarantees.

Verifiability: The transactions and interactions can be verified by the corresponding actors. This verification can be performed in real time.

Security: Only actors with clearance can add content or access to it. A robust

underlying consensus mechanism ensures that the transactions are signed

in a cryptographically secure way.

The first step involves the case creation. In this regard, a case can be registered due to a citizen’s testimony (we include in this definition any individual that wants to report a crime) or directly by a prosecutor (or an investigator with enough clearance to open a case) who observed suspicious behaviour. Next, evidence is collected and analysed by using the appropriate forensic tools. The description of each action (e.g. storing an evidence and analysing an evidence) may have a description file associated with JSON or CSV format, to ease further searches and classifications. We assume that secure and private storage is used to preserve the evidence, but other platforms such as cloud-based storage or decentralised storage systems such as the InterPlanetary File System (IPFS) can be used if data are properly protected/encrypted or other national IT-security guidelines. More concretely, depending on the approach selected by the investigators, the hashes of the evidence can point to an IPFS address and/or record the SHA-256 hash of the evidence. Therefore, our proposed system enables the extraction of the hashes of an investigation via a set of smart contract functions, enabling investigators to use such intelligence services (e.g. by using their APIs to upload the evidence or by a hash query) to retrieve additional intelligence. Finally, when the investigation concludes, all the data can be collected and presented in court. The aforementioned interactions are mapped into a smart contract and therefore stored permanently in the blockchain. The latter guarantees the verifiability of the investigation due to the blockchain’s immutability, as well as the preservation of the chain of custody. Therefore, the investigation can be audited to certify that any evidence was tampered during the investigation, guaranteeing the soundness of the different forensic procedures. In addition, our approach is designed to be accommodated and in other digital investigation contexts apart from embezzlement, enhancing its adaptability to internal audits.

In addition, the benefits go further beyond the provision of a solid proof in court, since the knowledge and evidence gathered in a case can be correlated in the future to reduce the time required to find a vulnerability or speed up cybercrime investigations. This is particularly relevant in the finance context, where all background and identification information can be stored. For instance, the Know your Customer (KYC) can be processed easier and faster during investigations and be secured against any internal fraudulent activities. On top of that, the application of smart contracts could prevent efforts for forgery and counterfeit documents. Therefore, the use of blockchain and its data mining capabilities will foster collaboration between different entities to share information and enable the early detection of embezzlement schemes. Moreover, evidence and reports can provide valuable input for the elaboration of AI models which increase the rate of embezzlement detection, since embezzlement schemes sometimes last for more than 5 years.

The key benefit from the managerial perspective of our proposed architecture is that it can eliminate operational risk and costs associated with

investigation procedures. In particular, it can accelerate investigations by facilitating the reconciliation of evidence in a verifiable and auditable manner,

increasing the possibility of identifying perpetrators and reducing fraud management and recovery costs. It enables senior management to have direct access to fraud risk exposure, reporting, and monitoring and gain a comprehensive view of all investigative information improving their decision-making analysis to impose sanctions and proceed to litigation actions. Based on the same methodological steps, the auditors can store their evidence from their audit trails to better monitor their follow-up actions. Thus, regular internal audits can rip the benefits of this architecture and deliver efficient outcomes to the management. Blockchain adoption might have a double effect on the internal audit department since it can reduce operational costs associated with the streamlining of auditing and investigative procedures.

Security problems such as identity theft and phishing are the major concern for both customers and merchants. Identity theft is the practice of stealing another

person’s identity for gaining access to his resource whereas phishing refers to the process of acquiring sensitive information by masquerading as a reputed entity. I’m proud to announce that Xaana is going to launch an E-payment system that provides an unrivalled security using visual and quantum cryptography. Proposed product will guarantee unconditional security over the traditional E-payment gateways. Visual cryptography hides the authentication details of customers by generating two shares for customer and bank respectively. Quantum cryptography secures the transmission of one time password. Encryption in quantum cryptography is done using quantum states called qubits. It is not possible to measure and clone the quantum states so that an eavesdropper cannot clone or measure the qubits. This technique requires two types of channels, quantum channel for secure transferring of key and classic channel for verification of key received.
‍

In the traditional E- payment system, after selecting the items from the E –shop portal, the customer submits his credit or debit card information for paying the merchant. The details submitted by the customer are verified by the bank and when the verification gets right the fund is transferred to the merchant's account. This payment system is not secure since any eavesdropper can act as a customer by hacking information submitted by the customer. In Xaana’s proposed payment system, a snapshot of text containing customer’s account number and debit and credit card information is taken. From the snapshot image two shares are generated using visual cryptography: one share will be in the hand of the customer and the other one will be in the bank's database. After that customer selects the desired items, the system transfers a list of items along with an encrypted account number to the bank. On receiving a list of items along with an encrypted account number, the bank generates a one time password and securely transfers it to the customer using quantum cryptography. After receiving one time password , steganography is performed by taking the customer's share as cover image and hidden information as one time password and stego image is passed to the bank. Bank extracts embedded one time password so that share and one time password gets separated. Then the Bank combines the customer's share with the bank's share and obtains account number and credit card details. Finally the bank validates the one time password and credit card details and if both verification gets right the fund is transferred to the merchant account number.

Sender can choose basis randomly for sending each bit of key to receiver

through a quantum channel. Receiver on receiving bits chooses a basis randomly to measure it and informs sender which basis he used through a public channel. On receiving the basis the sender informs the receiver when he has chosen the basis correctly. Keys consist of bits for which receiver has chosen the basis correctly. New keys consist of bits for which receiver has chosen the basis correctly. Subsets of the new key are compared in order to detect eavesdropping. Eavesdropping is not possible in this protocol since measurement of polarisation of photons using incorrect basis results in change of polarisation of photons. As a result, the receiver does not get the correct bit even after using the same basis as that of the sender and during comparison of a subset of keys, mismatch occurs indicating eavesdropping. If any of the verification gets wrong, the transaction fails.