ENG26SS103. The AI Agent Era: Accurate Books at AI Speed, presented by Puzzle (PRA, NAA)
Accounting firms today want three things: accurate books, a faster close, and more clients without more headcount. AI promises all three; however, for most firms, the promise hasn’t been delivered because the AI being sold to them was never built for accounting.
Everyone wants to close the books faster, and the promise of AI-driven productivity is everywhere. But for accountants, speed is a liability if it comes at the expense of accuracy. If you can’t trust the output, you can’t sign off on the financials.
In this session, Blake Oliver and Sasha Orloff demonstrate how firms are finally moving from using probabilistic automation to building on a deterministic foundation, where AI agents handle categorization and reconciliation continuously, but every transaction is governed by rules you set, and nothing posts to the general ledger without your explicit approval. You’ll learn the critical difference between bolted-on AI and an AI-native architecture built into the ledger, not on top of it, so every transaction is audit-ready from day one. Using Puzzle as a live case study, see how accountants become the architects and AI becomes the engine, keeping books current in real time and allowing you to catch errors as they happen. Leave this session with a practical framework to adopt AI that puts you in control and scales your firm’s capacity without scaling headcount.
Learning Objectives:
- Identify the structural differences between “bolted-on” AI and agent-native architecture to determine which approach provides the reliability required to deploy at scale.
- Recognize the technical requirements of an “audit-ready” general ledger (GL) to ensure that increased automation speed does not compromise financial accuracy or professional liability.
- Identify how GL-integrated AI agents transition the month-end close from a high-pressure batch process to a real-time review cycle, effectively doubling team productivity.
- Select a practical framework for implementing AI agents that reinforces, not replaces, professional judgment, applicable to any firm regardless of the tools currently in use.
