COASE.

Before a single figure is produced, an engagement demands a structure: which sections, in what order, supported by which exhibits, computed by which model. MIES holds that structure as a library of modules and sub-modules. From engagement type, scope, recipient and information level, the applicable modules assemble — mandatory, recommended, optional — into the report skeleton and the Excel files behind it. No text is generated. The professional populates the specific content.

Modular assembly of deliverable structures and supporting calculation files for M&A, valuation, financial modelling, FP&A, restructuring, project finance, transaction support, LBO, financing and debt advisory. Sector-specific variants.

Information memorandum, vendor and buyer due diligence, valuation report, fairness opinion, business plan document, independent business review, financing memorandum, restructuring plan, board pack, quarterly reporting. Standardisation of method across a practice. Onboarding of new professionals. Taking over an engagement in progress.

• Configuration — engagement type, scope of work, recipient (board, lender, investor, court, counterparty), information level, reference date, sector, framework
• Module library — sections and sub-sections as discrete modules, each with its purpose, the question it answers, the evidence it requires and the exhibits it carries
• Assembly logic — modules classified mandatory, recommended or optional against the configuration; every classification traceable to the rule and the standard behind it; the professional adds, removes and overrides, with the effect on the structure shown
• Dependencies — modules that require, exclude or feed one another; incompatible selections flagged at assembly, not at delivery
• Content guidance — per module: what must be stated, which evidence supports it, which methodological choices must be disclosed, which limitations declared; drafting guidance, never drafted conclusions
• Calculation architecture — Excel files assembled to match the selected modules: sheet structure, input-calculation-output separation, named ranges, formatting convention, documented assumption sheet
• Deterministic models — the calculation models each module requires, pre-wired into the files; formulas live, parameters open, conclusions left to the professional
• Consistency — figures flow from calculation files to exhibits; a change in an assumption propagates to every dependent exhibit
• Depth modulation — the same engagement type assembled at summary, standard or full depth; the module set scales accordingly
• Build sequence — order of compilation, dependencies between modules, checkpoints, information request list derived from the selected modules
• Completion tracking — what is populated, what is outstanding, what is inconsistent across the deliverable
• Variants — sector, size, jurisdiction, reporting framework; house modules and firm conventions added to the library
• Records — module set and overrides versioned; assumption log; decision log for the engagement file

Report skeleton with module-level guidance. Excel calculation files with live formulas. Exhibit set. Information request list. Build sequence. Engagement file index.

Assembles the structures the suite computes into: components from PROUVE, valuation in TANGE, plan in NERVI, debt in BOOLE, accounts from AALTO. Inherited models reviewed through PIANO. Drafting register via LOOS. Confidential material handled through ANDO.

On request.

A template shows where the content goes. A finished deliverable shows how deep it goes, how the exhibits are built, how the model behind them is wired. PROUVE is the populated side of the MIES corpus: complete reports and complete Excel models, worked end to end, usable on their own. The whole artefact, or the module alone.

Reports, Excel model files and Excel modules for M&A, valuation, financial modelling, FP&A, restructuring, project finance, transaction support, LBO, financing and debt advisory. Standalone use or integration into an existing model.

Setting the depth and structure of a deliverable before writing it. Starting a model from a working structure instead of a blank file. Adding a self-contained calculation block to an existing model. Benchmarking an internal deliverable against institutional depth. Training and internal reference. Second opinion on how a given analysis is normally built.

• Complete reports — worked deliverables by engagement family and situation, populated end to end: sections, exhibits, methodological disclosures, limitations and caveats as they appear in practice
• Complete models — Excel files matching those deliverables: input-calculation-output separation, named ranges, documented assumption sheet, live formulas throughout
• Modules — self-contained calculation blocks, each with its inputs, outputs, conventions and integration notes; usable without the parent model
• Explicit logic — no hidden macros, no hard-coded values inside formulas; every figure traceable to its input
• Documentation — technical specification per model and module; glossary; source register; stated caveats and known limitations
• Selection — by engagement family, situation, analysis or single calculation
• Integration — modules designed to be dropped into a third-party model; dependency and convention notes supplied
• Conventions — consistent structural, naming and formatting rules across the library
• Versioning — models and modules versioned; changes logged

Report files. Excel model files with live formulas. Excel modules. Technical specifications.

The populated side of the MIES corpus. Modules feed TANGE, NERVI and BOOLE. Accounts from AALTO. Third-party models reviewed through PIANO.

On request.

Valuations are rarely attacked on the number. They are attacked on the flows, the normalisations and the rate. TANGE builds the valuation from those three points outward: every choice sourced, quantified and traceable to a formula.

Enterprise and equity value of going concerns, listed and unlisted. Standalone entities and groups. Controlling and minority interests.

Buy-side and sell-side pricing. Fairness and expert opinions. IAS 36 impairment testing. Shareholder exit, withdrawal and squeeze-out. Restructuring plans. Litigation and arbitration. Review of third-party valuations. Negotiation preparation.

• Methods — DCF unlevered (FCFF/WACC), levered (FCFE/Ke), APV; economic profit, reconciled to DCF; sum-of-the-parts
• Multiples — trading and transaction; screened peer set with documented criteria; calendarisation, LTM/NTM, outlier treatment; regression-based value maps
• Horizon — two- and three-stage, fade to steady state; mid-year convention, stub period, valuation-date handling
• Terminal value — Gordon growth, value-driver formula (g/RONIC), exit-multiple cross-check; implied multiple, implied growth, TV weight
• Cash flows — normalisations itemised and quantified; IFRS 16 consistent across EBITDA, net debt and WACC; steady-state capex, D&A, NWC; tax-loss schedule
• Cost of capital — CAPM build-up: risk-free spot vs. normalised, ERP by source, bottom-up peer beta, Blume adjustment, Hamada / Harris-Pringle / Miles-Ezzell with debt beta, size and country premia, documented specific risk; synthetic-rating cost of debt; target capital structure; pre-tax rate derivation for IAS 36
• Currency and inflation — nominal/real consistency, inflation-differential and forward-rate approaches
• Bridge — net debt, debt-like items, leases, pension and severance obligations, minorities, associates, surplus assets, tax assets
• Equity claims — options, warrants, convertibles; treasury-stock method
• Adjustments — DLOM, control premium, minority and holding company discounts
• Sensitivity — real time: WACC × g, WACC × exit multiple, tornado, reverse DCF, scenario weighting
• Audit trail — linear calculation breakdown; every parameter with source, date and rationale; assumption log with versioning
• Challenge pack — assumptions exposed to dispute, expected questions, argued answers; drawn from proprietary deterministic sets

Excel model with live formulas. Valuation report. Methodology notes. Football field. Assumption log. Challenge pack.

Historical accounts from AALTO. Forecast from NERVI. Debt schedules from BOOLE into FCFE and bridge. Components from PROUVE. Report register via LOOS. Challenge-pack contextualisation through ANDO.

On request.

Most of the time spent on a business plan goes into building it, not into questioning it. NERVI inverts the ratio. Actuals in; a complete, integrated steady-growth model out, on day one. The professional works by difference, on the discontinuities that define the trajectory.

Integrated three-statement projections for operating companies and groups, across the corporate life cycle: expansion, development, maturity, decline, distress. Monthly to multi-year horizons.

Business plans for boards and lenders. Budget, rolling forecast and re-forecast. Financing applications. Vendor plans, buyer cases and independent business reviews. Turnaround plans and negotiated crisis settlement. Forward-looking monitoring of organisational arrangements. Impairment-test projections. Valuation inputs.

• Intake — historical accounts from AALTO, Excel or trial balance; monthly, quarterly, annual; reclassification; data-quality checks; normalisations
• Diagnostic — rule-based classification of life-cycle phase from growth, margins, capital intensity, leverage and cash conversion; sector profile; user override
• Baseline — integrated P&L, balance sheet and cash flow generated from actuals; drivers inferred from history; balanced by construction, no manual plug
• Techniques by phase — maturity: trend, mean reversion, operating leverage; development and expansion: price × volume, capacity, cohorts, unit economics, S-curve ramp-up; decline: run-off, cost stickiness; distress: 13-week cash flow, liquidity bridge, standstill and restructuring measures
• Revenue — price, volume, mix; by business line, customer, geography; backlog and pipeline conversion; seasonality
• Costs — fixed-variable split, inflation indexing, headcount-based personnel plan
• Working capital — DSO, DPO, DIO by driver; seasonality; factoring
• Capex — maintenance vs. expansion; depreciation by vintage; IFRS 16
• Financing — schedules from BOOLE; revolver logic, cash sweep, dividend policy, equity injections, covenants
• Tax — current and deferred; tax-loss carry-forward; interest-deductibility cap; country modules
• Work by difference — discontinuities as dated events with ramp-up and impact; every override logged against baseline; baseline-vs-current delta view
• Scenarios — base, downside, upside, stress; break-even; covenant-breach analysis; reverse stress test
• Sensitivity — real time on EBITDA, cash, leverage, DSCR and enterprise value
• Valuation link — TANGE valuation updated within the same process
• Monitoring — actual vs. plan variance; rolling re-forecast; 12-month forward DSCR
• Integrity — balance check, cash reconciliation, no hard-coded values in calculations, time-series consistency

Three-statement model. KPI set. Plan tables for the document. Sensitivity dashboard. Excel with live formulas.

Actuals from AALTO. Debt from BOOLE. Valuation in TANGE. Planning skeletons in MIES. Components shared with PROUVE. Plan narrative register via LOOS.

On request.

A debt schedule is a contract translated into cash flows. Standard tools translate the instalment; they lose the floor, the ratchet, the sweep and the way clauses interact. BOOLE's clause library was built by dissecting executed facility agreements: the terms in hand are configured clause by clause, and each one computes, period by period.

Debt instruments of corporate and structured finance: bank facilities, bonds, mezzanine and shareholder financing, leasing. Single position and entire debt structure. Euro and multi-currency.

Structuring of new financing. Term-sheet comparison and negotiation. Refinancing, amend-and-extend. Debt portfolio management and reporting. Amortised-cost accounting. Covenant monitoring. Debt module for business plans, LBO and project finance models. Independent recomputation of existing schedules.

• Instruments — term loans, RCF, capex and acquisition lines with availability period, bridge loans, bonds and minibonds, mezzanine, PIK and PIK-toggle notes, shareholder loans, leasing
• Configuration — facility set up from the term sheet: clause-level parameters, no formula writing; every computed figure linked back to the clause that produced it; templates by lender and facility type
• Repayment profiles — constant instalment, constant principal, bullet, balloon, sculpted to CFADS or target DSCR, custom; grace and pre-amortisation periods
• Rate mechanics — fixed, floating, switch; Euribor by tenor, €STR and SOFR compounded in arrears with lookback and observation shift; floor on index vs. floor on all-in rate, cap, collar; margin ratchet on leverage grid; step-up and step-down; sustainability-linked adjustment; default interest
• Conventions — ACT/360, ACT/365, 30/360, 30E/360, ACT/ACT; following and modified following, adjusted and unadjusted accrual; TARGET calendar; short and long stub periods; rounding rules
• Options and clauses — voluntary prepayment with break costs and make-whole; mandatory prepayment from cash sweep, excess cash flow and disposal proceeds; extension options; interest capitalisation; upfront, commitment, arrangement and agency fees; explicit order of clause interaction
• Rates — current fixings, market-implied forward curves, user-defined scenarios, parallel and non-parallel shifts
• Hedging — IRS, cap and collar overlay; hedged vs. unhedged debt service; hedge ratio over time
• Accounting — amortised cost and EIR with transaction costs; modification vs. extinguishment test; current and non-current split; accruals; journal-ready schedules
• Covenants — DSCR, LLCR, ICR, leverage, gearing; headroom at each testing date
• Portfolio — multiple facilities; consolidated debt service, maturity ladder, WAL, weighted average cost, fixed-floating mix
• Metrics — IRR, EIR, all-in cost, APR, WAL, duration
• Comparison — alternative terms side by side; differential cost over the life of the facility; refinancing break-even
• Replication — import of existing schedules; replication to the cent; differences itemised by period and by clause
• Audit trail — for each period: fixing, fixing date, day-count fraction, clause triggered

Schedules by facility and consolidated. Comparison report. Covenant headroom table. Amortised-cost schedules. Excel with live formulas. API.

Debt service into NERVI. Net debt and FCFE into TANGE. Financing, LBO and project finance skeletons in MIES. Components shared with PROUVE.

On request.

An XBRL instance is built for filing, not for analysis. Turning it into a working reclassification is the preliminary work no engagement pays for. AALTO does it as a verified chain: XBRL fact, checked QName, statutory statement line, management statement line. Every contribution keeps its context, QName, source line, multiplier and amount. Where the filing does not determine a figure, the field returns UNAVAILABLE and residuals stay visible — a statement that ties because a gap was filled with zero is worse than one that declares the gap.

Extraction and reclassification of filed financial statements into a standard financial reclassification. Does not reconstruct management accounting absent from the filing. Does not interpret the narrative notes.

Historical base for valuation and planning. Peer and competitor analysis. Credit and counterparty assessment. Portfolio monitoring. Due diligence data room processing. Multi-year time series. Batch processing of filings.

• Deterministic chain — XBRL fact → verified QName → statutory statement LINEA_ID → management statement line; candidate mappings never enter runtime
• Statutory reconstruction — statements S-01 and S-02 rebuilt before reclassification
• Traceability — each contribution retains context, QName, source line, multiplier and amount; every reclassified figure opens to its components
• Tie checks — reconciliation prevails over apparent completeness; undetermined items returned as UNAVAILABLE; residual amounts left visible, never absorbed
• Quality Engine — file validity, reconciliation and analytical granularity assessed separately; grade is a configured weighted average — not a probability, not a credit score
• Reclassification — standard reclassification formats; multi-year series with consistency checks across periods
• Frameworks — OIC, IFRS and US GAAP instances
• Single model — Excel and API derive from the same in-memory model

Reclassified management statements. Statutory statements S-01 and S-02. Contribution detail per line. Quality report. Excel. API.

Historical base for NERVI and TANGE. Accounts into MIES deliverables. Components shared with PROUVE.

On request.

A model shows its results. It does not show how they were produced, which cell silently overrides a formula, or what breaks if an input moves. PIANO reconstructs the architecture by reverse engineering the file: the dependency graph, the calculation paths from input to output, and the structural weaknesses along them. Findings are placed in their relational context, so the professional can judge materiality instead of reading a list. Two versions of the same file compare on their logic, not their numbers. The file is never modified.

Structural analysis of Excel models: architecture, dependencies, integrity and fragility. Not a review of the modelling assumptions, nor of the accounting or valuation methodology.

Taking over a model built by someone else. Due diligence on a target's financial model. Pre-delivery review of an internal model. Review of a bank's or advisor's model. Handover on staff departure. Deciding whether to repair or rebuild. Comparing two versions of the same model.

• Architecture map — sheets, blocks, roles (input, calculation, output, exhibit); flow of the model as built, against the flow it declares
• Dependency graph — precedents and dependents at cell, range, sheet and file level; interactive, filterable, reducible to a readable subgraph
• Calculation paths — full trace from any output back to its inputs; depth of dependency; longest chains
• Impact analysis — downstream reach of a given input or assumption: which outputs it feeds, through which paths, at what depth
• Formula inventory — distinct formula patterns, repetition, inconsistency within a range, volatile functions, array and legacy constructs, complexity per cell
• Integrity findings — hard-coded values inside formulas, broken and inconsistent ranges, circular references and iterative calculation, error values and their propagation, unit and sign inconsistencies, cross-sheet mismatches
• Structural findings — orphan sheets and dead cells, unused inputs, duplicated calculation logic, single points of failure, over-long chains, hidden sheets, rows and columns
• External surface — links to other workbooks and data sources, refresh dependencies, missing or stale sources
• Candidate overrides — manual values written over formula cells, isolated and listed with their location in the calculation path
• Named ranges and defined names — coverage, collisions, unused and broken names
• Version comparison — structural diff between two versions: formulas changed, logic rewritten, ranges moved, overrides introduced or removed; not a value-by-value comparison
• Macros — VBA presence, entry points, sheets and ranges written by code; declared as the limit of static analysis rather than analysed as formulas
• Protection and provenance — protected and locked areas, author and revision metadata, external references
• Findings register — each finding with location, relational context, severity and rationale; suppression with documented reason
• Scoring — structural quality by dimension, on configured weights; not a probability, not a judgment on the model's conclusions

Architecture map. Interactive dependency graph. Calculation path traces. Findings register. Version diff. Audit report. Excel export of findings.

Inherited models reviewed before adoption in MIES engagements. Structural standard shared with PROUVE. Third-party valuation models examined before review in TANGE. Debt schedules cross-checked against BOOLE.

On request.

Sending a client document to an external language model is a disclosure. ANDO puts a local step in front of it: sensitive elements are replaced by tokens, the mapping stays inside the working environment, and only the tokenised version leaves. When the answer comes back, ANDO restores the original references. Recognition is proposed and reviewed before anything is sent — a pseudonymisation tool that claims full automatic coverage cannot be relied on.

Preparation of confidential documents for processing by external language models, and local recomposition of the output. Not a redaction tool for publication, and not a substitute for the duty to assess whether a document may be processed externally at all.

Drafting and reviewing reports on client material. Analysing contracts, agreements and correspondence. Working on a target's documents during due diligence. Summarising data room material. Translation of confidential documents. Use of external models under professional secrecy or NDA.

• Deterministic detection — tax codes, VAT and company numbers, IBAN, national identifiers, email, telephone, addresses, URLs, dates, amounts: pattern-based, complete against the defined patterns
• Entity recognition — local NER model for legal entities, persons, brands, locations and roles; runs offline, no call leaves the machine; results presented as proposals with confidence, never applied silently
• Review before send — every proposed substitution listed in context, accepted, rejected or edited; nothing is transmitted before review; entities missed by detection added manually
• Consistency — one entity, one token, across the whole document and across every document in the same session; relationships, roles and cross-references preserved so the model can still reason on the text
• Token design — type-preserving tokens (a company reads as a company, an amount as an amount); numeric scaling and date shifting with a preserved arithmetic relation, so figures stay usable without being real
• Contextual residual — quasi-identifiers flagged after substitution: sector, location, size, market position, distinctive facts that can re-identify a subject with every name removed; the judgment is left to the professional, the exposure is not left hidden
• Structure retention — formatting, tables, numbering and document structure preserved, so the returned output recomposes without manual rework
• Mapping — held in memory for the session; persisted only on explicit choice, encrypted, stored locally, with defined retention and destruction
• Recomposition — original references restored in the returned output; tokens left unresolved by the model reported rather than silently dropped
• Custom dictionaries — client, project and firm-specific terms; do-not-send terms that block transmission outright
• Batch — multiple documents processed under a single consistent mapping
• Audit trail — what was substituted, what was reviewed, what was sent, what returned; log per session, exportable for the engagement file
• Sealed boundary — no telemetry, no network access in the processing step; operation verifiable with the machine offline

Tokenised document ready for transmission. Substitution register. Recomposed output. Session log.

Placed in front of any external model used on MIES deliverables, PROUVE material or TANGE challenge packs. Register conventions shared with LOOS.

On request.

A general language model writes fluently, and gets it wrong. It calls a finding an insight, states as verified what was only represented, and shifts register between one section and the next. In a professional report, those are not stylistic slips: "indicates", "suggests" and "we were unable to verify" carry different degrees of commitment. LOOS supplies the lexical and structural apparatus of institutional reporting, and checks the draft against it. Form only. The analysis remains the professional's.

Terminology, formulation of judgments and register consistency in professional documents. Assisted drafting through an external model, or direct use as a reference by the professional. Not a content generator, not a fact checker.

Drafting valuation reports, due diligence reports, fairness opinions, business plan documents, independent business reviews, memoranda, board packs. Review of a draft before delivery. Alignment of documents written by several hands. Translation of professional documents between Italian and English. Training and internal reference. Definition of a house register.

• Lexicon — professional terminology by domain, with the distinctions a general model collapses: accounting, valuation, transactional and legal terms kept apart where they differ in meaning
• Certainty gradation — formulations graded by degree of commitment: verified, represented, estimated, not verifiable; the wording that corresponds to each, and the wording that overstates it
• Judgment formulation — how a finding, a limitation, a caveat, a qualification and a recommendation are stated; what must accompany each to stand on its own
• Structural formulations — openings, transitions, cross-references, methodological disclosures, limitation and reliance paragraphs, standard closings, by document type
• Register — formality, distance, use of the first person, active and passive voice, conventions consistent across sections and documents
• Prescriptive rules — preferred, discouraged and prohibited terms, with rationale; commercial and promotional language flagged out of professional documents; hedging and overstatement both identified
• Retrieval — semantic search over the structure base: the formulation that fits the context, with its usage conditions
• Draft validation — a draft checked against the lexicon and the rules: terminological inconsistency, unwarranted certainty, register drift, unsupported claims, internal contradictions in wording; each flag with the rule behind it
• Bilingual equivalence — Italian and English terms paired as professional equivalents, not literal translations; the cases where no equivalence exists are marked
• House layer — firm-specific terms, conventions and prohibitions added on top of the base, with precedence over it
• LLM integration — supplied to the external model as retrieved context and as constraints, so the register is instructed rather than requested
• Provenance — each entry with its documentary type of origin and usage conditions; abstracted linguistic structures, never source text

Retrieved formulations with usage conditions. Validation report on the draft. Term and equivalence lists. House register.

Drafting register for MIES deliverables and PROUVE reports. Terminology for TANGE valuation reports and NERVI plan narratives. Used through ANDO where the draft contains confidential material.

On request.