Ep 772: AI You Can Trust: When Good Enough Isn’t Actually Good Enough
May 7, 2026
AI Summary
5 min readJeremiah Edwards, head of Sage AI, discusses building trustworthy AI for finance professionals on the Everyday AI Podcast. Drawing from Sage's accounting software for small and medium businesses (SMBs), he explains Sage Copilot's new finance intelligence agent, which speeds up closing books, tracks budgets, and answers business questions. The conversation emphasizes that rapid advances in agentic AI demand rigorous checks for correctness and explainability, especially where numbers drive decisions.
Why Explainability Matters in Finance
Finance leaders face high accountability—tax filings and audits fall on CFOs, not AI models. A PwC study cited shows 71% reject AI without explanations. Edwards stresses that "close is not good enough"; outputs must be fully auditable, showing data sources, API calls, and reasoning steps. This design principle in Sage Copilot builds trust by making every agent action traceable, countering the rush for more outputs without verifying quality.
Continue reading the full summary in the app — free to try.
Read Full Summary →Free • No credit card required
What you'll learn
- 1 (00:53) **AI Trust and Auditing** - Host questions if "good enough" AI suffices amid rapid agentic growth
- 2 (01:39) **Building Trust in AI for Finance** - Focus on explainability and correctness gaps in AI tools
- 3 (02:45) **Sage Company Overview** - Sage provides accounting software for SMBs worldwide
- 4 (03:12) **Sage Copilot New Capabilities** - UI for generative/agentic AI, including finance intelligence agent
- 5 (04:25) **Balancing Agentic Growth with Accuracy** - Beyond models: needs agent harness, tools, data connections like MCP/A2A
- 6 (06:19) **LLM Math Capabilities Update** - LLMs trained on text, prone to internet errors; need tools like calculators
- 7 (07:51) **PwC Partnership on Explainability** - 71% of finance leaders reject non-explainable AI
+ Full timestamped outline available in the app
Show Notes
What good is AI if you can't trust it?
Everyone's racing to do more, build more, ship more agents. But if you're scanning an AI output and say it looks good enough, that isn't actually good enough.
Especially when it comes to your company's finances.
If you're a CFO, a finance leader, or running an SMB where one wrong number sends everything sideways, having trust in your AI is paramount.
On today's show, we're getting into why the model alone won't save you, where finance pros should actually be spending their time now, and why most companies are thinking about AI risk backwards.
Helping us break it down is Jeremiah Edwards, Head of Sage AI, from Sage Future last week.
Newsletter: Sign up for our free daily newsletter
More on this Episode: Episode Page
Today's Episode on LinkedIn: Thoughts on this? Join the convo on LinkedIn and connect with other AI leaders.
Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineup
Website: YourEverydayAI.com
Email The Show: [email protected]
Connect with Jordan on LinkedIn
Topics Covered in This Episode:
- AI Trust and Explainability in Finance
- Sage Copilot's New Agentic AI Features
- Financial Accuracy and Model Risk Management
- Chain of Thought Reasoning in AI Outputs
- Explainability and Auditability for CFO Confidence
- Human Agency and AI Automation in Accounting
- Shadow IT Risks and Model Integration
- Risk Appetite for SMB Finance AI Adoption
- Outlier Detection and Anomaly Monitoring with AI
- Future-Proofing Financial Processes with AI
Timestamps:
00:00 Discussing AI trust and auditing
05:17 AI models and math capabilities update
07:50 Importance of AI explainability
12:39 AI in finance and time-saving
13:52 Optimizing time with agentic AI
18:18 Integrating AI into teams
21:54 Using AI in different roles
24:37 Adopting AI with real value
26:59 Wrapping up and subscribing
Keywords:
AI you can trust, trustworthy AI, audit your AI, AI explainability, explai
More from this podcast
Everyday AI Podcast – An AI and ChatGPT Podcast →