SFNet’s Reimagined: AI & Automation in Secured Finance conference was held June 2 at the offices of Winston Taylor. This one-day sold out event was designed for lenders, fintech innovators, and technology leaders seeking tangible strategies to modernize asset-based lending and factoring operations. The conference featured expert panelists and hands-on sessions and a Vendor Showcase. Attendees walked away with actionable insights into AI-powered fraud detection, API-driven collateral monitoring, digital borrower integrations, and advanced portfolio analytics.

Similar to other technological advances, the key conference themes were that AI is shifting, but not replacing, the labor environment and having a Human-in-the-loop (HITL) is still necessary and continues to be more effective in using AI. A lot of training, testing and benchmark data are still needed to use AI to the fullest. In the secured finance industry, some companies have AI fully integrated into their systems while others are just starting and neither approach is incorrect.

Embedded Monitoring in ABL & Factoring, sponsored by CORA, kicked off the conference, and looked at how the secured finance industry is moving toward daily, API-driven asset monitoring, enabling secured finance teams to ingest real-time data directly from client bank accounts, e-commerce platforms, logistics systems, and AR/AP systems. API and MCP-enabled asset verification pushes daily AR, inventory, and receivables data straight into firm dashboards. Panelists included Alejandro Hernandez, chief technology officer, SLR Digital Finance; Pieter de Kok, head of Clients & Data, Finterface; Albert Spada, executive vice president and head of Business Capital, M&T Bank; and Jeff Liguori, co-founder, CORA. Nancy Lee, co-founder and CEO, ABLSoft, served as moderator.

Panelists discussed the pros and cons of continuous monitoring, the state of industry, adoption, and challenges. Borrowers and lenders are used to getting data on a fixed frequency, such as once per month or week. However, having data given at the same moment in time, for all asset classes, is a relevant indicator of timing fraud and the risk of fraud is significantly decreased. Receiving inventory lists once per month, for example, and accounts receivable lists once per week, often results in mismatches and potential double financing since reports are not given at the same time. AI technology offers quicker ways to flag the anomalies found in data vs. going through reports manually.

Companies have done a better job of imposing reporting discipline on their customers and requiring a system to be set up at closing is the best way to get ongoing discipline in the system. Panelists agreed it is harder to change the way reporting is given once the customer is onboarded.

Embedded monitoring is predicted to take hold the fastest in accounts receivable, quickly followed by inventory valuation. Products that can revalue inventory on a real-time basis are going to be a powerful solution versus receiving annual or semi-annual appraisals.

Commercial banks, who may operate in a more bureaucratic space, may be better suited with a gradual rollout. The ABL space, which often has more reporting and analysis involved, and more areas to ingest data from, may still be a few years away from embedded monitoring.

Transparency will continue to be a challenge, and a human is needed to check automation. Lenders should also account for changes on the borrower’s side, such as new warehouses, products or supplies. Getting a customer to trust lenders with their credentials and allow direct access to their systems is another challenge that needs proper framing, such as fewer audits.

The Fraud & Credit Risk Resilience panel, sponsored by AIO Logic, provided actionable insights for ABL lenders and factors on fortifying fraud controls in an increasingly digital landscape. Drawing from real-world cases like First Brands and recent market trends, the discussion covered evolving fraud tactics including fake entities, fabricated collateral, and circular payments. Panelists also shared key takeaways from SFNet’s Fraud Task Force Report, "Systemic Resilience in Secured Finance."

Robert Bowles, founder, executive director, Bluewater Transaction Advisors LLC served as moderator. Panelists were Paul Bower, director, FGI Tech, Dan Ennis, partner, Parker, Hudson, Rainer & Dobbs LLP and Daniel Pravich, co-founder, ExamFlow.

A stacked, three-layer defense model, consisting of a calculation layer, or a deterministic software layer, the AI or a detection layer, which can detect anomalies when going through large amounts of data, and the human, or judgment, layer, was discussed as the best approach against fraud.

Panelists stressed that fraud often happens when lenders become complacent with a client. They may skim documents, not follow the initial checklist that was first used and rush past red flags over time. The AI detection layer can capture anomalies that a human may miss, especially with large amounts of data.

AI, however, cannot replace a field examiner who has face-to-face conversations with borrowers. Panelists agreed that as the human judgment layer, field examiners are not going away any time soon. In five years, AI can help do the work before the judgment, such as reconciliations, so field exams switch from being rushed engagements to providing more value from the get-go.

The panel also examined the anatomy of a modern fraud, which was divided into billing integrity and inventory “re-aging”, as well as AI as the Shield in transforming monitoring and transitioning from reactive, points-in-time audits, to continuous, embedded monitoring architectures. In the end, AI can automate analytical grunt work, ingesting thousands of invoices, reconcile ledgers and track serial numbers to detect “zombie inventory.”

According to SFNet’s 2026 Blue Ribbon Task Force Report, 62 percent of lenders need better fraud detection. Legal structuring complements, but does not substitute for, collateral diligence. There needs to be a call to action to adopt real-time monitoring tools, strengthen cross-lender communication and invest in training to spot changing fraud.

The Conference featured two Vendor Showcase and Breakout Sessions featuring technology and service providers including Codix, FGI and XEN whose solutions directly support a particular focus on embedded monitoring, fraud and credit risk management. The second showcase highlighted vendors ExamFlow, LoanWatch and RelPro, whose products help lenders and factors modernize how they monitor collateral, manage cash, detect early warning signals, and improve decision-making through real-time data, automation, and AI-enabled analytics:

Lunch featured keynote speaker Nick Pericle, founder of Tenexity, who discussed "Translating AI from HYPE to HOW for the Industries That Move the Economy", who did hands-on demonstrations using readily available AI tools to illustrate ways secured lenders can prompt and create workflows, agents and apps geared for the ABL and factoring industries. Pericle discussed which AI tools are better at which tasks, such as Copilot working best with emails, and Claude currently being ranked as best overall. Pericle demonstrated the creation of a Borrowing Base app using an AI called Replit as well as document summarization demonstrations. He noted treating the AI tool as a teammate and collaborative partner vs. a tool where one action is taken, can chang the outputs and what people can achieve.

Executive-Level Strategy for ABLs and Factors: Data-Driven Decision Making for Portfolio Health, sponsored by RelPro, evaluated where AI, automation, and advanced analytics can create real business value to asset-based lenders and factors. This discussion brought together factoring and ABL leaders who have or are actively seeking implementing technology-enabled strategies, alongside software and AI specialists building tools for private credit, factoring, and secured finance applications.

Bill Elliott, president, Accounts Receivable Financing Group, First Business Bank was the moderator, and panelists were Jack Bowen, client partner, KUNGFU.AI; Lin Chua, president & CFOO, InterNex Capital; Nate Eisenberg, sr. product manager for Data & Analytics, Allvue Systems; and Raul Esqueda, president, 1st Commercial Credit, LLC.

The panel began by advising what lenders should start with when deciding to convert from legacy systems to AI. They should look across their systems for rich data, precedents, and ask what problems they are trying to solve. On the factoring side, look at the inbound of documents, package it to where you are auditing and get to where you are funding it. The second would be the underwriting process, automated UCC searches and background checks, and the third agent would be for remittance and posting. One panelist mentioned MCA lenders have been using AI for several years and are processing approximately 12,400 applications a day using AI in the application and preliminary evaluation stage. Their average loan size is moving up to $4 million and creeping into the ABL space.

InterNex’s approach was to build out their AI systems from scratch and Chua cautioned that implementing AI is not a static exercise, it is ongoing for learning capabilities and spending. When the company began in 2015, Chua said they were deliberate in bringing in professionals experienced with tech, making AI seem like any other technological advancement rollout.

Generative AI is difficult to audit as they are probabilistic, not deterministic, systems producing new results with every prompt. There are better tools out there, such as gradient boosting tree, a known statistical framework for data, is more favored for regulatory bodies.

Providing guardrails for AI agents was recommended, so that it doesn’t answer with whatever it wants. Verify the models, and it can take 120 days to compare models with A/B testing.

Innovation That Integrates with Legacy Systems, sponsored by Iridium Credit, closed out the panels for the day. Panelists explored how asset-based lenders and factors can modernize operations without undertaking a full system replacement. Many institutions still rely on legacy core platforms, separate tools for borrowing base certificates, field exams, credit files, and monitoring, as well as manual, email-driven workflows. Panelists discussed practical approaches for introducing modern technology into these environments.

Errin Glasgow, president, Cascade Credit Services, LLC served as moderator and panelists included Mark Gorman, SVP, EDI coordinator, Rosenthal Capital Group; Jeff Carlson, co-founder, QBIX Analytics & Loanwatch; Billy Quinn, general manager, CODIX USA; and Alexander Kayfetz-Gaum, senior vice president, XEN.

Panelists began by explaining that new technology doesn’t need to replace legacy software, but can instead complement it. Rip and replace carries internal and external risks. When considering updating a system, consider time, cost and opportunity and ask how much it would take to grade versus replace. Technologies have a lifecycle and maintenance is required. Organizations tend to get stuck with getting people onboard.

Once an organization does decide Glasglow asked how they cut through the noise and decide on the right system. Having a goal and starting with the right problems, such as needing to get more information easily and quickly, as opposed to solutions, is the first step. Instead of looking for a vendor, look for a partner instead. Ask prospective vendors to try options out before committing, check references and ask if a dedicated human is available throughout the journey.

Panelists advised having a proper project plan, including buy-in from different entities, and realize that the best laid plans can go awry and it’s best to have a fallback plan when considering implementing AI.

The best practices for introducing new tech into legacy environments include real world testing. One panelist advised assembling people from day-to-day usage to join testing environments. Doing yesterday’s work as a way to test and see if the same results are delivered is another method.

Panelists concluded the panel by sharing their final thoughts on what the future looks like in an AI-integrated world: There is a lot of opportunity, but the key is to continue moving; there is competitive risk to inaction;  technology success depends on the people as much as the systems; and implementing AI is not a one-time project; it is an ongoing valuation.

The conference closed out with a Networking Reception, sponsored by AIO Logic.