Everyday AI Podcast – An AI and ChatGPT Podcast Ep 541: AI & Trust: When 98% accuracy won't cut it and how Sage can fix it
39 snips
Jun 6, 2025 
Join Aaron Harris, CTO of Sage, a leading firm in cloud accounting, as he dives into the vital connection between AI and financial accuracy. They discuss how even a 98% accuracy rate can shake trust in finances. Aaron highlights Sage's innovative approaches, including their 7-billion parameter model and AI Copilot, designed to boost precision. The conversation also covers the importance of transparency in AI labeling and the collaboration needed for industry-wide trust. Get ready to reimagine how finance teams can confidently integrate AI into their workflows!
AI Snips
Chapters
Transcript
Episode notes
Near-Perfect Accuracy in Finance
- Finance requires near-perfect accuracy because even one penny off can lead to loss of trust.
- CFOs and finance teams will hunt down any discrepancy in the books to maintain credibility.
Use AI Judiciously
- Use AI only when traditional development can't solve the problem better.
- For tasks like math, pair AI with specialized tools like calculators rather than relying on AI creativity.
Domain-Specific Model Fine-Tuning
- Fine-tuning smaller, domain-specific AI models drastically improves accuracy and trustworthiness.
- Training on accounting-specific data and company APIs yields a model with expert financial knowledge, reducing hallucinations.