The method behind the numbers.

This page explains how the MurphFin AI calculation engine works and why the financial plans it produces are structurally different from those generated by pre-filled templates or by generative artificial intelligence models. Our goal is to produce documents that hold up in front of a credit committee, not documents that merely look professional.

Deterministic engine, not generative

MurphFin AI is a proprietary financial calculation engine. It is not a language model, it is not a neural network, it is not a black box. It is a mathematical engine that applies explicit financial formulas to the data entered by the user.

Given the same inputs, the engine always produces the same output. Results do not depend on probabilities, temperatures or stochastic parameters. Every figure in the business plan — in the financial statements, the charts and the narrative sections — is calculated by this engine, not by a language model.

The engine is developed in-house by Analytiko S.r.l. (formerly Fisit S.r.l.), a company based in Trento, Italy, that carries out applied research and develops software platforms and AI systems for highly regulated sectors, delivered as SaaS, within the NVIDIA Inception programme, and undergoes systematic audit cycles across all supported grant programmes, with cross-checks between income statement, cash flow, balance sheet, financial ratios, incentive calculation and taxes.

Transparent, editable drivers

The financial plan does not start from templates or industry averages. It starts from the company's actual data, entered by the user through a guided 9-step wizard.

• Volumes, unit prices and product mix — defined by the user
• Fixed and variable cost structure — itemised line by line
• Investment plan by expenditure category — with differentiated depreciation schedules
• Headcount and labour cost — calculated on actual CCNL (Italian national collective agreement) pay scales (8 agreements, level by level)
• Days sales outstanding, days payable outstanding, days inventory on hand — parameters entered by the user
• Funding sources, interest rates and loan amortisation schedules

Every parameter is visible, editable and flows directly into the financial statements. There are no hidden assumptions. The professional has full control over the drivers and can verify at any time how input data translates into financial output.

Complete financial statements

The engine automatically generates four integrated statements over a 5-year horizon:

The four statements are mutually consistent: the cash flow statement is derived from the income statement and the change in NWC, and the balance sheet closes with the cash position calculated from the cash flow statement. If one statement changes, all the others update automatically.

Bank DSCR and covenants

The DSCR (Debt Service Coverage Ratio) is calculated using the standard banking formula, in line with EBA and Bank of Italy guidelines. It is not a simplified DSCR computed on net profit: it is built on actual operating cash flow, net of taxes and of the change in net working capital.

The calculation is performed year by year, not as an average over the period. Debt service includes the instalments of subsidised loans and bank facilities included in the plan.

Beyond the DSCR, the engine automatically calculates and tests a set of standard bank covenants — including leverage ratios, liquidity ratios and debt coverage — with a traffic-light system that highlights critical values year by year.

The aim is not to hide problems: it is to surface the weak points before submission, so that the professional can act on the drivers and build a sustainable plan.

Net working capital

Working capital is one of the most underestimated elements in financial planning. A plan with growing revenues but unmodelled working capital can show positive profits and negative cash — a signal that any credit analyst recognises immediately.

MurphFin models net working capital (NWC) on three actual parameters entered by the user: average days sales outstanding, average days payable outstanding and days inventory on hand. From these three values the engine calculates trade receivables, trade payables and inventory for each year of the plan.

The annual change in NWC flows directly into the cash flow statement and into the DSCR calculation. The cash flow statement also includes a monthly projection with the actual timing lag of collections and payments, based on real days — not on proxies or flat-rate estimates.

Scenarios and sensitivity analysis

A financial plan with a single scenario is not a plan: it is a statement of intent. MurphFin generates three parallel scenarios — optimistic, realistic and pessimistic — calculated over 5 years with user-defined parametric variations on revenues, costs and investments.

The scenarios are not linear. They do not apply a constant growth rate: the user defines the variation parameters for each scenario, and the engine recalculates the entire financial plan (income statement, cash flow statement, balance sheet, ratios) for each one.

The sensitivity analysis measures the impact of changes in key drivers — revenues, variable costs, fixed costs — on the plan's net profit. This allows the professional to quantify the plan's resilience and to identify its points of vulnerability before the credit analyst does.

Narrative: a tool, not the final product

The descriptive sections of the business plan (project description, market analysis, operating plan, innovation, employment impact) are generated by an external language model. This model is used exclusively for the narrative component.

Every figure quoted in the text is calculated by the MurphFin AI deterministic engine, not by the language model. The model does not invent numbers: it takes them from the financial plan that has already been calculated.

The generated narrative is a working draft, not a sealed output. The user can rewrite, supplement or replace any section. The built-in AI Coach provides operational guidance item by item — no theory, only concrete actions based on the plan's data.

A business plan that is "too perfect" is not credible. That is why MurphFin produces a solid structure that the professional tailors with their own expertise and knowledge of the specific case.

Open, verifiable export

The financial plan can be exported as a professional PDF and in Excel format. The export is not a closed document: the professional can open the model, verify the data, add further analysis and present it under their own signature.

MurphFin is a working tool for professionals, not a substitute for financial analysis. It speeds up the construction of the plan, ensures consistency across statements and automates repetitive calculations — but responsibility for the plan remains with the professional who signs and presents it.

Historical financial statements and grounded projections

For companies with a track record, MurphFin allows historical financial statements to be imported from Excel (.xlsx) or XBRL (.xbrl) files. The parser automatically recognises and maps income statement and balance sheet items, including headcount and share capital.

Historical financial statements feed the comparison with the projections: the professional can verify the consistency between past performance and the forecast plan, and the credit analyst finds the historical trend data they expect.

Automatic validation and consistency

The engine includes a validation system that compares the data entered against reference ranges by expenditure type, sector and geographical area. If costs or revenues are out of scale compared with industry averages, the system flags it.

The incentive calculation is calibrated on the specific constraints of each grant programme: eligible expenditure categories, investment thresholds, non-repayable grant percentages and access conditions. If the parameters fall outside the constraints of the selected programme, the system warns the user before the plan is generated.

The calculation engine undergoes systematic automated testing that verifies cross-consistency between all statements for every supported grant programme.