Most agent courses stop when the demo runs. This one goes on through what decides whether an agent survives a risk review.

✓ Real market data  – Labs run on SEC filings and market data. An agent that only works on a curated demo prompt fails visibly in the debrief.
✓ Cost budgeted early  – Every agent is instrumented for token cost and latency. A crew that wins at ten times the price is reported as such.
✓ Adversarial by default – Document poisoning, prompt injection and red teaming are built into the labs, not mentioned as a caveat.
✓ Governance that maps – Kill-switches, audit trails and an EU AI Act mapping a second-line risk function can actually work with.

---

What you take away

Capstone & deliverables

Sessions 1 to 6 each end with a component. Session 7 assembles them into a finance-sector agent of your choosing, with the evidence to prove it meets its targets.

✓ Six lab systems and an evaluated capstone — seven in all
✓ Cost, latency and failure instrumentation you can reuse
✓ An audit and governance pack mapped to the EU AI Act obligations in force as of October 2026

Assessment & completion

✓ Practical labs — 60%. Six graded lab systems, one from each of sessions one to six.
✓ Final project — 30%. The capstone agent against its safety and performance targets.
✓ Participation — 10%. Engagement and peer review of other crews’ agents.
✓ Successful completion requires a final weighted mark of at least 70/100, attendance at six of the seven live sessions, and submission of all seven systems.
✓ Marks and certification are awarded individually.
✓ AIFI Professional Certificate upon successful completion.