What Makes an AI System "Agentic"?
The foundational distinction that changes everything: from systems that advise to systems that act. Why this is not a technical upgrade but a categorical shift in governance requirements.
Read Brief 1 →The Business Analytics Institute offers a wide range of training, coaching and consulting services to help management improve their ability to take tough decisions.
BAI Summer School 2026 · Technical Briefs
Ten issues. Every other Tuesday. Free. A deep dive that closes the gap between what universities teach and what European banks are deploying — before 2 August 2026 changes everything.
European banks are deploying agentic AI systems right now — multi-agent credit underwriting, autonomous AML monitoring, AI-driven onboarding. But most of the professionals and students who will govern, build, or work alongside these systems have never been taught what actually separates a production agentic system from a Jupyter notebook exercise.
Convergence Technical Briefs are a ten-issue, twice-monthly reading programme drawn directly from the introductory module of the BAI Summer School 2026. Nine briefs publish before the Summer School opens on 1 July. A tenth — The Governance Paradox — arrives on 21 July, eleven days after the programme closes.
The briefs are preparation. The programme is practice. The capstone is reflection.
"The question for banks is no longer whether to adopt agentic AI. It is whether they understand what they are adopting — and whether their governance structures are ready for systems that act, not merely advise."
BAI Summer School 20264,000+Convergence™ Technical Briefs
subscribers4 ECTSZHAW
accreditation
Documented deployments from UBS, JPMorgan, Goldman Sachs, Wells Fargo, and Morgan Stanley — with outcomes, not vendor promises.
LangGraph, Model Context Protocol, OpenAI Agents SDK, ReAct loops — the tools banks deploy, not the Jupyter notebooks universities teach.
EU AI Act Articles 9–15, GDPR Article 22, DORA — mapped to specific technical decisions, not legal summaries.
Five questions per brief with guidance notes — calibrated to reveal what you don't yet know, not just confirm what you do.
Arrive at Anglet already fluent in the vocabulary the programme's case discussions will assume — not building it in real time.
Brief 10 publishes on 21 July — after the programme closes — designed to land differently once you've lived through the case discussions.
Pre-programme deep dive · Briefs 1–9 · 17 March – 7 July 2026
The foundational distinction that changes everything: from systems that advise to systems that act. Why this is not a technical upgrade but a categorical shift in governance requirements.
Jensen and Meckling's 1976 framework meets LLM deployment. When your agent is an algorithm, misaligned incentives become misspecified objectives — consequences play out at machine speed before anyone notices.
An evidence-based tour of production deployments reshaping global banking — with documented outcomes, not vendor promises. JPMorgan, Goldman Sachs, Wells Fargo, Capital One, Morgan Stanley.
How production agentic systems are actually constructed. The five-layer architecture, three orchestration patterns, and the tool stack that separates what banks deploy from what universities teach.
A deep technical case study reconstructing Wells Fargo's multi-agent loan re-underwriting system — annotated code patterns, governance decisions embedded in each layer, lessons for any production deployment.
£4 million in damage. No technical violations. Every decision within delegated authority. The governance failure scenario that reveals why existing frameworks were designed for the wrong kind of AI.
Agentic systems can cause material financial damage in 48 hours. The shift from governance by audit to governance by design — with the specific mechanisms that close the gap.
EU AI Act Articles 9–15 mapped to specific compliance obligations. GDPR Article 22. DORA. The accountability gap left by the withdrawn AI Liability Directive.
JPMorgan's framing of the new division of labour. The three human capabilities agentic AI is elevating to strategic importance — and why traditional banking training programmes are not building them.
Post-programme capstone · Brief 10 · 21 July 2026
The central strategic tension of 2026: the speed and scale creating competitive advantage are precisely the properties making governance harder. Designed to land after the programme — when Anglet becomes the frame through which its argument is read.
Why Brief 10 publishes on 21 July
Brief 10 is not pre-reading. It is a reflective capstone — written to land differently once you have spent ten days building and governing agentic systems in practice. Ten issues of preparation. Ten days of programme. One capstone for reflection.
Free · Every other Tuesday · 17 March – 21 July
Subscribe to receive all ten briefs directly. Free to all. No paywall. No prerequisites. Just the gap, closing — issue by issue, every other Tuesday.
The briefs are also published at www.baisummer.com and remain available throughout the programme period.
GDPR compliant·Unsubscribe anytime·No spam
Free · Ten issues · Twice monthly
You'll receive The Convergence Technical Briefs every other Tuesday, starting 17 March 2026.
Check your inbox for a confirmation email.
www.baisummer.com · info@baisummer.com