Finance in Flux

Key Concepts

• Data Snipper: Software company specializing in automating data collection, analysis, extraction, and matching for audit and finance teams.
• AI Augmentation: Using Artificial Intelligence to enhance, rather than fully replace, human capabilities in accounting and finance.
• Vertical AI: AI solutions tailored to specific industries and their unique workflows and regulatory requirements.
• Agentic Features: AI capabilities that can perform a series of consecutive, repetitive tasks, acting as automated assistants.
• Audit Trail: A chronological record of financial transactions and system activities, crucial for regulatory compliance and trust.
• Hallucination (AI): The generation of incorrect or fabricated information by an AI model.
• IFRS (International Financial Reporting Standards): A set of accounting standards used in many countries.
• GAAP (Generally Accepted Accounting Principles): A common set of accounting principles, standards, and procedures by governments and business.
• Disclosure Testing: A process in auditing that involves reviewing financial statements for required disclosures.
• Operational Risk: The risk of loss resulting from inadequate or failed internal processes, people, and systems or from external events.

Data Snipper: Automating Audit and Finance with AI

Data Snipper is an Amsterdam-headquartered software company focused on revolutionizing audit and finance processes. The company addresses the long-standing challenge of data management, which involves collecting data from stakeholders, analyzing it, extracting relevant information, matching disparate data points, and maintaining a robust audit trail. Data Snipper leverages AI to automate these tasks, aiming to improve efficiency and accuracy for audit and finance teams globally.

The Accounting Profession: A Dwindling Supply and Increasing Demand

The transcript highlights a critical trend in the accounting industry: a significant shortage of qualified accountants, particularly CPAs. This shortage is attributed to several factors:

• Declining interest in accounting studies: Fewer students are pursuing accounting degrees.
• Shift to tech and AI firms: Graduates with accounting degrees are increasingly opting for careers in technology and AI companies.
• Rising regulatory requirements: The number of regulations in accounting and finance continues to grow, increasing the complexity and workload.

This creates a scenario of dwindling supply and increasing demand, leading to concerns about job loss due to AI being misplaced. Instead, the speakers express optimism that AI, like Data Snipper, will augment accountants, making the profession more engaging and attractive by automating tedious tasks and allowing professionals to focus on higher-value activities.

Barclays' Experience with Data Snipper

Paul, representing Barclays, shared his organization's experience with Data Snipper. He described the "soul-destroying work" involved in ensuring financial statements accurately tied back to source data. Data Snipper has addressed this pain point by:

• Automating manual processes: Significantly reducing the manual effort required from financial control teams.
• Enhancing control environment: Creating a more robust system for financial reporting.
• Reducing operating risk: Dramatically decreasing the risk associated with quarter-end processes.

Barclays views AI as a transformative technology with immense financial opportunities, aligning with their mantra of being "better, simpler, and more balanced." They utilize AI for productivity, efficiency, revenue generation, and improving their control environment, with a long history of AI application in areas like fraud detection and customer service.

Navigating Highly Regulated Industries

Selling technology into highly regulated industries like finance presents unique challenges. Companies are primarily concerned with accuracy and trust, needing to prove to regulators that they are not making mistakes and have comprehensive audit trails.

Data Snipper's approach to overcoming this has been to:

• Prioritize accuracy and trust: Building software that minimizes reliance on AI "hallucination" or subjective judgment.
• Tie back to evidence: Ensuring all AI-driven insights are directly linked to verifiable data.
• Incorporate auditability: Building audit features directly into the software to track actions, time saved, and ROI.

Paul from Barclays echoed this, stating that Data Snipper helps address the control environment and reduces operational risk, which is of immense benefit to stakeholders like regulators and the board.

Competition and Verticalization in AI

The AI landscape is becoming increasingly competitive, with large frontier model companies like Anthropic (Claude for Financial Services) and OpenAI developing specialized financial services products. Vija, CEO of Data Snipper, argues that workflow and use cases matter more than just the underlying AI model. He believes that in complex, regulated industries, vertical solutions will ultimately win because they possess a deeper understanding of industry-specific workflows and regulatory requirements. Data Snipper's strategy involves embedding itself alongside customers to develop products tailored to their exact needs, giving them a "right to win" in the long term.

Barclays, while embracing AI, maintains strict guidelines for its use, including model validation, data assessment, compliance, and human interaction. They have strong governance structures in place, chaired by the company's COOs, to oversee AI projects.

The Rise of AI Agents

The discussion touched upon the concept of AI agents, which are effective for performing a series of consecutive, repetitive tasks. Vija highlighted disclosure testing as a prime example where Data Snipper's agents excel. Unlike generic models that might simply answer questions, Data Snipper's agents are designed for complex decision-making ("should I answer these questions?") and detective work, incorporating specific regulatory frameworks like GAAP and IFRS. This "purpose-built" approach, he contends, makes their agents significantly more effective than generic models in regulated environments.

Barclays is also exploring agents for tasks like anomaly identification and narrative generation, and is considering their application in legal departments for contract review and in investment banking for due diligence and regulatory filings.

Government Sector Adoption

Data Snipper has also found significant traction in the government sector, as governments are often the largest auditors globally. Audit organizations within every government department are tasked with ensuring tax dollars are allocated appropriately. With governments facing vast amounts of data, Data Snipper offers a solution for increased efficiency and accuracy in their audits. The UK Parliament has publicly acknowledged Data Snipper, citing a potential for a "3x efficiency" gain for the UK audit office. Data Snipper has dozens of government customers in the US at state, county, and city levels.

Data Training and Security

A crucial aspect of building AI solutions is data acquisition. Data Snipper emphasizes that they do not train on customer data due to the highly confidential nature of pre-published financial information. Instead, they focus on building products iteratively alongside their customers, ensuring the functionality and logic are correct without compromising client data security. This commitment to data security is fundamental to their trust with large organizations and governments.

Collaboration with Startups at Barclays

Barclays actively seeks out emerging technology companies and innovation. They have an accelerator program and an experimentation hub that allows them to work with startups in quarantined testing environments using real data, fostering a "fast fail" mindset. Paul, as Chairman of Investment Banking, is personally involved in identifying partners like Data Snipper, seeing opportunities for co-development, product roadmap integration, and providing banking capabilities to support their capital cycle.

Upskilling and the Future of Accounting Training

The conversation addressed the potential for AI tools to upskill individuals, enabling them to perform tasks that previously required advanced certifications. While regulations vary by country (e.g., the UK does not require certification to practice as an auditor), the speakers believe AI will attract more people to the profession by shifting focus from "grunt work" to more engaging aspects like risk assessment, detective work, and strategic advising.

Drawing parallels to the adoption of QuickBooks, which didn't eliminate bookkeepers but rather changed how they worked, the speakers anticipate a similar evolution in audit and accounting with AI. Big Four firms like PwC are already considering how to upskill their teams, aiming for entry-level hires to function at a managerial level by managing AI agents.

The Role of the Human in the Loop

Regarding the auditability and traceability of AI agents, the discussion highlighted that while the "human in the loop" can sometimes be a problem, it remains essential in audit. Determining the right controls is an art that requires human judgment, understanding the client's business, identifying vulnerabilities, and assessing risk. While agents can efficiently process data and verify compliance with controls, the fundamental definition and implementation of those controls still rely on human expertise.