CPA firms grapple with safe AI deployment beyond basic chatbots

TLDW: General chatbots fail in accounting firms due to context loss between conversations and unclear security protocols for sensitive data uploads.

Key points:

• Consumer AI chatbots lack persistent context, forcing firms to rebuild explanations for every new query
• Security remains ambiguous: each file upload presents judgment calls about compliance and confidentiality
• Firms are adopting unapproved AI tools despite risks, signaling demand for safer alternatives
• Agentic AI (with memory and state management) addresses context persistence but requires careful deployment
• Firms need governance frameworks before scaling AI beyond initial experiments