S12 E25: From Processing $500B at Intuit to Building the Fraud-Proof Ledger: Eradicating Financial Misstatement with Ahikam Kaufman, Co-Founder & CEO of Safebooks AI

Episode Summary

Executive Summary: Ahikam Kaufman, CEO of SafeBooks AI, explains how his finance background led him to build an agentic platform that automates office-of-the-CFO work by connecting data across systems and reducing repetitive close, reconciliation, and audit tasks. He argues finance is ripe for AI because of data complexity, talent shortages, and growing acceptance of automation, and says the company’s key advantage is a data context graph that improves accuracy and trust.

Main Topics: Origin of SafeBooks AI (Priority: 5/5): Kaufman traces the idea to his years in finance and fintech, where closing books and handling regulatory complexity created obvious opportunities for automation. Agentic automation for finance operations (Priority: 5/5): SafeBooks aims to emulate human finance work across structured data workflows, especially repetitive tasks in close, reconciliation, and reporting. Data context graph as the foundation (Priority: 5/5): The company spent nearly two years building a financial data context graph that links CRM, billing, contracts, ERP, banks, payroll, and other systems to give AI full context and reduce hallucinations. Product roadmap shaped by customer pain (Priority: 4/5): After working on real customer data, the team noticed repeated complaints about slow close cycles and focused the product on close acceleration and execution rather than generic finance tooling. Team, culture, and security (Priority: 4/5): Kaufman emphasizes hiring data experts and AI-first people, building trust with customers, and making security a top-level engineering principle because finance data is highly sensitive. Future of the office of the CFO (Priority: 5/5): He predicts major automation over the next three years in finance functions like AP, AR, controller, SEC reporting, FP&A, and audits, while human judgment remains essential for governance and accounting decisions. Advice for founders (Priority: 3/5): Kaufman advises entrepreneurs to treat investors and teams as partners, hire for culture as well as skill, move fast, and build around a real business pain rather than technology alone.

Key Arguments: Finance operations are full of repetitive, cross-system work that AI can do faster and with sufficient accuracy. The hardest finance problems are not isolated tasks but reconciling data across multiple systems with integrity and auditability. A proprietary data context graph is essential for trustworthy AI in the office of the CFO because it gives models the right context and reduces hallucinations. Generic close-management software helps manage the process, but SafeBooks focuses on executing the work itself. Customer discovery revealed a consistent pain point: close takes too long because of manual validation, documentation, and analytics. AI adoption in finance is accelerating because the technology has reached a level where it can meet enterprise expectations for quality and confidence. Finance is facing a talent shortage and a massive opportunity for automation to do more with less. Security and trust are central because finance customers are handling sensitive, regulated data and will not adopt tools they do not trust. The future of finance tools is augmentation: systems of record remain, but they will be powered by much more automation and agentic workflows. Founders should solve a real business problem first and only then automate or productize it.

Data Points: Companies founded by Kaufman: 3 - He says he has started three companies. Company exit: 1 exit to Intuitive - Kaufman notes one prior company exit to Intuitive. Public companies worked in: 4 - He references working in finance across four public companies. MVP build time: almost 2 years - SafeBooks spent nearly two years building the first financial data context graph. AI maturity comparison: 6 to 8 months - He says finance-grade AI accuracy and quality have only been sufficient for the last six to eight months. Future transformation window: 3 years - He predicts the office of the CFO will look very different within three years.

Pivotal Quotes: "We developed an agentic platform that can totally emulate most of the work done by a human being, replacing repetitive specific work around structured and unstructured data." — Ahikam Kaufman: Explaining what SafeBooks does and why finance automation is feasible "We spent a lot of time, almost two years, building the first what we call financial data context graph... we have to provide AI the right infrastructure and foundation to make sure it doesn't hallucinate and it doesn't make any errors." — Ahikam Kaufman: Describing the product foundation and differentiation "I think entering into the office of the CFO is in its infancy right now. I think we're just getting started." — Ahikam Kaufman: On the near-term future of AI in finance operations

Implications: Finance is becoming an AI-first operations domain. Tools that unify data context, preserve trust, and automate execution will likely reshape close, audit, and reporting workflows, while human finance professionals shift toward judgment and governance.