AI for Leveraged Finance
Financial modelling and loan and bond documentation analysis. AI can read a credit agreement and build a leveraged finance model; doing either reliably is a skill, and it is the subject of this masterclass. Wednesday 20 and Wednesday 27 January 2027, 18:00–21:00 Madrid.
Reliable output from AI in leveraged finance.
The masterclass sets out the working method behind that reliability: how the work is directed, what the model is given to read, how its output is verified, and where experience and judgement have to take over. It is applied to the two activities that absorb most of the time in leveraged finance, financial modelling and the analysis of loan and bond documentation.
Working method behind LevFin Academy.
LevFin Academy’s models and covenant materials were built with AI as a working companion: the judgement remained human and the execution was delegated. Using AI well in that work has taken far more than a single prompt.
Good output comes from drafting, challenging, correcting and rebuilding, often many times over.
As a conversation grows, the model’s performance deteriorates; past a certain length it is, in effect, exhausted, and the work has to restart from a clean summary.
Specialist agents multiply what one person can do: one reads the restricted payments covenant, another the debt incurrence provisions, and a third checks both against the definitions.
A PDF is expensive and inefficient for a model to process; the same document in markdown is lighter, searchable and easy to split by clause.
A knowledge base, such as an Obsidian vault written in markdown, gives the model a memory that no chat window has.
Finance is not code. There is rarely a test that proves an answer right: a forecast or the reading of a covenant is judgement, built on knowledge kept inside banks and funds rather than in public data.
Two sessions, one working method.
Session 1 establishes the method and applies it to loan and bond documentation; Session 2 applies the same method to building and auditing a leveraged finance model. Every worked example uses public documents.
Iteration as the unit of work. Drafting, challenging, correcting and rebuilding an output until it can be defended.
Context management. Why output deteriorates in long conversations, when to restart, and how to hand the work over to a fresh conversation through a clean summary.
Projects, skills, plugins and connectors. What each one extends, and when each one is worth the set-up.
Formats. Why a PDF is an expensive input, and how to convert a credit agreement into markdown split by clause, with definitions and cross-references preserved.
Knowledge base. An Obsidian vault written in markdown as persistent memory across conversations and projects.
Specialist agents. How to divide a task, brief each agent and reconcile their outputs.
Confidentiality. Firm policies, approved tools, and what never goes into a public model.
Definitions first. Consolidated EBITDA, Consolidated Net Income and the add-backs that drive every ratio in the document.
Specialist agents by provision. Restricted payments (builder, general and ratio baskets); debt incurrence (ratio debt, incremental and free-and-clear capacity, most-favoured-nation protection); permitted investments and reallocation between baskets; asset sales and the excess cash flow sweep; change of control and portability; events of default.
Verification. Clause-level citations, a checking agent, and reconciliation against a credit agreement checklist.
Loan and bond documents. Loan Market Association (LMA) language with US high-yield structure in recent European credit agreements; senior facilities agreements and indentures compared.
Worked examples on public documents. Dole plc’s credit agreement of May 2025 and Nomad Foods’ senior facilities agreement as restated in October 2025.
Where judgement takes over. Interpretation, market practice, and the questions the documents leave open.
Layered method. The IFRS presentation of the statements, their interconnections, the schedules behind each line, the independent accounts and their drivers, then the applications, with a clear line between what AI builds and what the analyst decides.
Specialist agents by area. Revenue and working capital; property, plant and equipment and intangibles; leases under IFRS 16; tax and loss carry-forwards; provisions and personnel costs; the debt schedule.
Building in Excel with AI. Formula construction, consistency across periods, and integrity checks on the balance sheet, the cash reconciliation and circular references.
Auditing a model with AI. Tracing every figure to a source, a link or a driver, and the error patterns that AI-built schedules typically contain.
Applications. The LBO, the debt schedule and returns; add-on acquisitions, repricings, refinancings and dividend recapitalisations; covenant capacity calculated from the documentation analysis of Session 1.
Where judgement takes over. Driver selection, downside cases framed in rating terms, and the decisions a sponsor actually takes.
Both sessions run live online with questions and discussion throughout, and both are recorded. Participants receive the recordings, the public documents of the worked examples in markdown, the agent briefs and checklists used in the sessions, and the model files of Session 2. Education and analytics only; not legal, credit or investment advice.
18:00–21:00 Madrid, 17:00–20:00 London, 12:00–15:00 New York. Both sessions run on the advertised dates.
Early rate for the first fifteen seats · €300 thereafter.
Seats and team editions.
Seats are booked and paid online. Prices are inclusive of applicable taxes, and an invoice is issued after payment.
For the first fifteen seats. Once they are taken, the early-rate link closes automatically and the standard rate applies.
One team of up to ten participants. Two live sessions of three hours on dates agreed with the team, with time set aside for the team’s own workflows and questions. The dates are agreed after booking.
Full refund if it’s not for you after Session 1. Invoices to a company name and VAT number can be requested at checkout. For a purchase order or any other question: federico.etchelecu@levfinacademy.com.
Built for practitioners.
Leveraged and acquisition finance bankers, private credit and direct lending teams, debt advisers, private equity deal teams, high-yield, leveraged loan and CLO credit analysts, and banking and finance lawyers. Working knowledge of leveraged finance and Excel is assumed, together with access to at least one AI assistant; the method is tool-agnostic.
Federico Etchelecu has spent more than twenty years in leveraged finance and investment banking, including sixteen years at Banco Santander, most recently as Executive Director in Leveraged Finance. He is the author of The LevFin Book and the Leveraged Finance Covenants Bible. About the instructor