What this course covers
Most people working in finance have heard about AI agents. Fewer understand what they actually do - or why the gap between what vendors promise and what businesses actually get is so wide.
This mini-course cuts through the noise. In 12 minutes you'll get a clear picture of what agents are, why the standard shortcuts fail, and what the right architecture looks like - using the Self-Service AI framework built and taught at LDI.
-
01
~2 min
Are agents overhyped?
A precise definition of what an AI agent actually does: perceive inputs, reason, act, and remember. And an honest answer to whether the capabilities match the hype.
-
02
~2 min
Why agents matter for finance teams
Four specific reasons agents create value in finance: deeper analysis across structured and unstructured data, faster adaptation, time savings, and better data governance.
-
03
~3 min
The three challenges with Gen AI for business
Why personal-productivity AI tools don't transfer cleanly to business process automation: scattered data, non-determinism, and business context that can't be handed over through prompts.
-
04
~3 min
The Self-Service AI framework
How to build an agent like hiring a team: a job role (markdown instruction file), skills (markdown knowledge files), tools (MCP servers), and the business data and context that makes it reliable.
-
05
~2 min
Five things to take away - and how to start
The five-step implementation path: data fundamentals, semantic models, agentic workflow design, building with human oversight, and testing in parallel with your existing process.
The five principles you'll leave with
- 01 Agents are powerful - but they don't work because of clever prompts alone. They work when they're built on the right structure and architecture.
- 02 Gen AI by itself is not enough for business processes. Business needs consistency, accountability, and control.
- 03 The best way to think about agents is to think about them like a team. Define the role, give them the right skills, connect them to the right tools.
- 04 Data is the foundation for everything. If the data is messy or fragmented, the agent's outputs will be too.
- 05 Start small - but start properly. Test in parallel. Trust is not built by assumption. It is built through results, consistency and time.